Coding, Modulation, and Link Protocol (CMLP) Study

Home , Turbo code

National Aeronautics and Space Administration (NASA) Space Communication and Navigation (SCaN)

Report

January 10, 2008

Les Deutsch Jet Propulsion Laboratory

and

Frank Stocklin Goddard Space Flight Center

Co-leaders Coding, Modulation, and Link Protocol Study Report

Table of Contents

1 Executive Summary 1 1.1 Results Summary 2

2 Charter of CMLP Study 7

3 CMLP Team Membership 8

4 Introduction 9 4.1 SCaN Architecture 9 4.2 Need for CMLP Study 13 4.3 CMLP Study Scope 13 4.4 CMLP Study Plan 14 4.5 CMLP Study Process 17 4.6 Plan for International Participation 18

5 NASA Links 20 5.1 NASA Space Network (SN) 20 5.2 NASA Ground Network (GN) 20 5.3 NASA Deep Space Network (DSN) 20 5.4 NASA Mars Network 20 5.5 Lunar and Constellation Links 21 5.6 Multiple Access Regimes 28 5.6.1 Near Earth Relay (NER) Space Network (SN) 28 5.6.2 Near Earth Ground Network 29 5.6.3 Lunar DTE/DFE 29 5.6.4 Lunar Relay 29 5.6.5 Deep Space DTE/DFE 29 5.6.6 Mars Relay 30

6 Current CMLP Schemes 31 6.1 Space Network (SN) Schemes 31 6.2 Ground Network (GN) Schemes 35

ii Coding, Modulation, and Link Protocol Study Report

6.3 Deep Space Network (DSN) Schemes 37 6.4 Mars Network 38

7 CMLP Schemes 39 7.1 Modulation 39 7.1.1 PSK 39 7.1.2 PCM/PSK/PM 41 7.1.3 PCM/PM/NRZ or Bi-phase 42 7.1.4 FSK and GMSK 45 7.1.5 QAM 45 7.2 Codes 46 7.2.1 Classic Codes 50 7.2.2 Modern Codes 55 7.3 Multiple Access Schemes 60 7.3.1 General Schemes 60 7.3.2 Time Sharing Schemes 61 7.3.3 Frequency Sharing Schemes 62 7.3.4 Direct Sequence Techniques 63 7.3.5 Random Access Methods 65 7.3.6 Hybrid Methods 68 7.4 List of Link Protocol Attributes 70 7.4.1 Background: Data Link Layer concepts, services, and functions. 70 7.4.2 Data Transfer: Transfer variable-sized service data units (SDUs) over serial links. 72 7.4.3 Data Transfer: Provide Segmentation and Reassembly. 72 7.4.4 Data Transfer: Provide fill data when required by Physical Layer; synchronization 73 7.4.5 Data Transfer: Provide Link Layer encapsulation and addressing 74 7.4.6 Data Transfer: Provide compatibility with multiple network layer protocols 74 7.4.7 Data Transfer: Minimize overhead (impact on throughput/utilization) 74 7.4.8 Data Transfer: Minimize impact on coding and lower layers 74

iii Coding, Modulation, and Link Protocol Study Report

7.4.9 Reliability and Quality of Service (QoS): Support class of service capability at Link Layer 75 7.4.10 Reliability and Quality of Service (QoS): Provide strong error detection capability at Link Layer 76 7.4.11 Reliability and Quality of Service (QoS): Provide error correction via automatic retransmission 76 7.4.12 Reliability and Quality of Service (QoS): Support rich Link Layer metrics for accountability 77 7.4.13 Channel Access and Usage: Operate over a shared channel: Virtual Channels 78 7.4.14 Channel Access and Usage: Operate over a shared channel: Medium Access Control (MAC) 78 7.4.15 Channel Access and Usage: Provide link establishment (hailing) 80 7.4.16 Channel Access and Usage: Provide channel management and link adaptation 81 7.4.17 Link Layer Security 81 7.4.18 Link Layer Functionality by Link Type 81

8 Figures of Merit (FOMs) 86

9 Navigation Considerations 88 9.1 Navigation within the Space Communication Architecture 88 9.2 Requirements and Methods 89 9.2.1 Analysis of the Navigation Requirements 92

10 Spectrum Constraints 101 10.1 Background 102 10.2 General Constraints 103 10.2.1 Spectrum Efficiency 103 10.2.2 Power Flux Density on Earth Constraints Imposed by the ITU-R Radio Regulations and ITU-R Recommendations 104 10.2.3 PFD Limit Relief for LEO and Data Relay Satellites 110 10.2.4 Other ITU-R Recommendations 111 10.2.5 Constraints Imposed By the Space frequency Coordination Group (SFCG) 112

iv Coding, Modulation, and Link Protocol Study Report

10.2.6 Other Relevant Emission Masks 120 10.3 Conclusions 120

11 CMLP Selection 122 11.1 Process 122 11.1.1 Inputs 122 11.1.2 Down-select Process 124 11.1.3 Outputs 125 11.2 Code and Modulation Interdependence 125 11.3 Modulations 127 11.4 Codes 147 11.4.1 Constraints Relevant to Code Selection 147 11.4.2 Down-select Procedure 148 11.4.3 Initial Code Down-selections 149 11.4.4 FOM Analysis 153 11.4.5 Final Code Selections 156 11.5 Multiple Access 158 11.5.1 Down-select Procedure 158 11.5.2 Initial MA Down-selection 161 11.5.3 Near Earth Scenarios – Final Selection Analysis 162 11.5.4 Lunar Scenarios 164 11.5.5 Code Division Multiple Access for Lunar Missions 168 11.5.6 GMSK/PN Multiple Access for Lunar Missions 185 11.5.7 FOM Analysis of Lunar Scenarios 192 11.5.8 Final MA Selections 201 11.5.9 Mars 203 11.5.10 FOM Analysis for Mars Scenarios 204 11.5.11 Future Work 205 11.6 Interaction of Link Protocols with Lower Layers 207 11.6.1 Link Layer ARQ 207 11.6.2 Virtual Channels vs. Physical Channels 209

v Coding, Modulation, and Link Protocol Study Report

11.6.3 Link Layer Metadata 210 11.6.4 Dynamic Access 210 11.7 Interrelation of Link-Layer Protocols to Upper Layers 211

12 Final Recommendations 212 12.1 Transition considerations 212 12.2 Schedule 213

13 Conclusions 216

14 Future Work 217

15 Acronyms 218

16 Appendices 223 Appendix 1: Performance Estimates for Two-Dimensional Coded Modulations 223 Appendix 2: GN and SN Modulation Downselect Workbook 223 Appendix 3: Near-Earth Constellation Modulation Downselect Workbook 223 Appendix 4: Category B Modulation Downselect Workbook 223 Appendix 5: Coding FOM Analysis Workbook 223 Appendix 6: Comparison of C2 and TPC 223 Appendix 7: Ranging with GMSK 223

17 References 224

vi Coding, Modulation, and Link Protocol Study Report

1 Executive Summary

The Coding, Modulation, and Link Protocol (CMLP) Study was chartered by NASA’s Space Communication and Navigation (SCaN) office to fill in some technical detail to the recently-adopted SCaN Architecture for future space communications. In particular, the CMLP study team was to recommend coding, modulation, link protocol attributes, and multiple access communication schemes to be used in each of the links defined in the architecture.

Only radio links were considered in the CMLP study. Although the SCaN architecture included consideration of optical communications, any practical operational use is far enough in the future, and too loosely defined, to permit any meaningful CMLP recommendations at this time.

The CMLP team only considered link protocols to the extent that they have a significant impact on the choice of codes, modulations, and multiple access schemes. An- other SCaN team is chartered with recommending actual link protocols. That team will use the CMLP recommendations on link protocol attributes as one of their inputs.

The CMLP team developed a catalog of links in the SCaN architecture through the year 2030 together with whatever is currently known about the eventual communications requirements of the corresponding missions. The team next created catalogs of all known and reasonable codes, modulations, link protocol attributes, and multiple access schemes. These catalogs, which were reviewed by a set of non-NASA US experts, formed the set from which subsequent down-selections were accomplished.

A set of “figures of merit” (FOMs) was developed to reflect performance, cost, and other constraints that would bias selection of the best schemes on a link. The FOMs were reviewed both by the outside US experts and by the SCaN Space Communica- tions Architecture Working Group (SCAWG).

The links in the SCaN architecture were classified into a smaller number of distinct “representative” links in order to reduce the amount of analysis needed. For each of these link classes, the CMLP team performed an initial down-select based on very high-level considerations. The resulting set of “reasonable” options for each link class was reviewed by a set of internal NASA experts.

Finally, the team performed detailed analysis on the remaining options for each link class with respect to the FOMs. The resulting recommendations are presented in Sec- tion 11.

1 Coding, Modulation, and Link Protocol Study Report

1.1 Results Summary

Table 1-1, Table 1-2, Table 1-3, and Table 1-4 provide an overview of the recommended coding, modulation, multiple access, and link protocol techniques. It should be noted that legacy modulation and coding techniques are recommended. While these legacy techniques may not be as powerful as other recommended techniques, they are, however, deeply embedded in the current infrastructure and will probably remain in use for some time. A notional transition plan is provided in Section 12.1 that will attempt to show an evolution but it is of course very subject to available funding and should be considered in that context.

Although the team has recommended TDRSS-style CMDA as the multiple access scheme for the Earth relay scenario, we were not able to come up with a single recommendation for the lunar direct-to-Earth scenario. All but two of the candidate schemes were eliminated – but both the CDMA or GMSK/PN schemes remain as possible recommendations. There were two obstacles to deciding on a single recommendation.

First, although CDMA works just fine in the nearer-term lunar scenarios, there may be cases in the further term where CDMA as it is currently configured may be over- whelmed. There are several possible solutions to this, such as obtaining additional bandwidth, using polarization diversity, and interference mitigation.

Second, although the GMSK/PN scheme may be more attractive when there are large numbers of channels, it is less technically mature and would require further work to bring it to the desired maturity level.

Since both schemes work in the near-term, the team proposes that NASA do to the additional technical work on the GMSK/PN scheme and reach a consensus on this within two years. If this time frame were too late (e.g. NASA’s Constellation Pro- gram is driving things faster than that) then we would have to recommend CDMA with the understanding that this could change in the out years.

In any case, the team recommends that

1) NASA continue to hone lunar scenarios, 2) NASA invests in a technology development effort for the GMSK/PN scheme, and 3) NASA invests in a technology development effort for CDMA “crowded signal scenarios”.

2 Coding, Modulation, and Link Protocol Study Report

Network Recommended Typical Application Parameter Technique  For low data rate, severely power-constrained links or links CCSDS turbo codes (r which have little spectral containment requirements = 1/6, 1/4, 1/3)  Typical application may be a small Mars surface platform with DTE/DFE link  Code family with general applicability to most links CCSDS AR4JA LDPC  Envisioned future replacement to traditional convolutional codes (r = 1/2, 2/3, and concatenated codes 4/5)  Offers superior coding gain over traditional codes  For links which are power-constrained and spectrum- CCSDS C2 LDPC constrained code (r = 7/8)  Typical application may be high data rate CAT A missions  Mid-Transition: General applicability to most links except spectrum-constrained links  Post-Transition: For severely latency-constrained links only Convolutional codes  Coding Offers superior heritage and reliability  Typical application may be a LEO mission’s TT&C link and the high rate science link  R-S and BCH have general applicability to most links, especially as outer codes  Need for an outer code is diminished with planned migra- Legacy Reed- tion to LDPC codes as noted above Solomon, BCH  Typical application may be a mission which launches prior to the demonstrated operational readiness of LDPCC- capable NASA infrastructure  For links which are not power-constrained  Although a mission may not be power constrained, consideration should be given to use of a code because of the ben- Uncoded efits to power flux density and resiliency to distortions  Typical application may be an X-band LEO mission down- linking at a very high data rate to the NASA GN Table 1-1: Recommended Coding Schemes

3 Coding, Modulation, and Link Protocol Study Report

Network Recommended Typical Application Parameter Technique For links to/from NASA ground infrastructure in which the residual carrier signal structure is the de facto mode of operation (e.g., the NASA GN) PCM/PSK/PM,  For low power links which cannot reliably ensure carrier PCM/PM lock with a suppressed carrier signal structure  Typical application may be a LEO mission’s S-band TT&C link to the GN OQPSK/PM  General applicability to most links except very high data rate links (where Trellis receivers and space-qualified digital modulators may not be practical) Precoded GMSK  Typical application may be spectrum-constrained S-band Modulation TT&C links via the SN or DFE/DTE  For very high data rate links in which filtered OQPSK/PM and Precoded GMSK hardware may not be feasible OQPSK (SQPSK)  Typical application may be ultra-high data rate science data link via the SN 650 MHz Ka-band channel  8PSK is recommended for links which are so severely spectrally constrained that 4PSK is not feasible  16QAM is recommended for Category B missions which are 8-PSK and 16-QAM so severely constrained that 4PSK and 8PSK are not feasible  Typical 8PSK application may be ultra-high data rate science data link via the SN 225 MHz Ku-band channel Table 1-2: Recommended Modulation Schemes

4 Coding, Modulation, and Link Protocol Study Report

Network Recommended Typical Application Parameter Technique

For SN forward or return S-band multiple access, use CDMA CDMA with spread spectrum PN codes through a multiple access DSSS phased array antenna

For SN forward or return Ku- or Ka-band multiple access, use Multiple Ac- FDMA an FDMA-type scheme using scheduled services through mul- cess tiple single access antennas with unique frequency slots and/or (Earth) spatial separation for each user

For GN S-/X- or Ka-Band multiple access, use scheduled ser- FDMA vices through multiple ground antennas which themselves employ multiple frequency bands

CDMA For lunar DTE/DFE GT forward or return S-Band multiple ac- DSSS cess, use CDMA with spread spectrum PN codes or GMSK with or PN code ranging through a ground antenna which provides FDMA multiple access GMSK/PN RECOMMENDATION UNDER STUDY

CDMA For Lunar Relay Satellite forward or return S-band multiple DSSS access, use CDMA with spread spectrum PN codes or GMSK or Multiple Ac- with PN code ranging through a multiple access relay antenna FDMA cess GMSK/PN RECOMMENDATION UNDER STUDY (Moon)

For lunar DTE/DFE GT forward or return Ka-Band multiple FDMA access, use scheduled services through ground antennas which provide multiple access

For Lunar Relay Satellite forward or return Ka-Band multiple FDMA access, use scheduled services through a multiple access relay antenna

Table 1-3: Recommended Multiple Access Schemes

5 Coding, Modulation, and Link Protocol Study Report

Network Recommended Typical Application Parameter Technique  The NASA GN and SN expect to continue, at the very least, to support a baseband data service which effectively implements no link protocol capability but does enable missions to operate any link protocol on top of the baseband data ser- None vice Link Protocol  This is a legacy technique which will continue to be used by existing NASA missions and “private” user missions but will not be available for new, future NASA missions. TBD TBD TBD Table 1-4: Recommended Link Protocol Attributes

6 Coding, Modulation, and Link Protocol Study Report

2 Charter of CMLP Study

Plug a hole that exists in the recommended NASA Space Communication and Navi- gation (SCaN) Architecture between the spectrum defined for each link and the network protocols.

Recommend and justify link designs for the SCaN Architecture.

Provide guidance on link design to the builders of NASA’s communications infrastructure, the developers of NASA’s spacecraft technology, the developers of NASA’s future mission concepts, and to the NASA Standards Program.

Identify key required NASA communications and navigation investments.

Engage NASA’s potential international partners in order to capture good ideas, build advocacy for international standards, and build good will.

7 Coding, Modulation, and Link Protocol Study Report

3 CMLP Team Membership

The CMLP Study Team is made up of NASA communications and navigation experts for various NASA centers. Key support contractors are also included in order to capture all the necessary expertise.

CMLP Study Co-leads: Les Deutsch (JPL), Frank Stocklin (GSFC)

Systems Engineering Subteam Lead: Gary Noreen (JPL) Coding & Modulation Subteam Lead: Jon Hamkins (JPL) Multiple Access Subteam Lead: Dave Zillig (GSFC) Link Protocol Subteam Lead: Loren Clare (JPL)

List of all team members:

Monty Andro (GRC) Alina Bedrossian (JPL) Dan Brandel (HQ) Loren Clare (JPL) Les Deutsch (JPL) Dariush Divsalar (JPL) Sam Dolinar (JPL) Pat Eblen (NASA HQ) Wai Fong (GSFC) Jay Gao (JPL) Jon Hamkins (JPL) Dave Israel (GSFC) Dennis Lee (JPL) Peter Militch Bob Nelson Gary Noreen (Cx) Richard Orr (Sa-Tel) Chitra Patel (GSFC) Fabrizio Pollara (JPL) Frank Stocklin (GSFC) Tudor Stoenescu (JPL) Scott Sands (GRC) Victor Sank (GSFC) Len Schuchman (Sa-Tel) John Wesdock (GSFC) Dave Zillig (GSFC)

8 Coding, Modulation, and Link Protocol Study Report

4 Introduction

In 2005, NASA created the Space Communications Architecture Working Group (SCAWG) to define a vision for space communications at the agency for the next 20 years. NASA had just embarked on its Vision for Space Exploration, including plans for a return of humans to the Moon and then to Mars. Such a large undertaking required coordinated planning, including communications and navigation support.

Since there was no central space communication organization at NASA at that time, the SCAWG comprised representatives from many NASA Directorates and centers.

The SCAWG was successful in developing a vision and architecture for NASA’s future space communications. NASA adopted the SCAWG-recommended architec- turei in principle in 2006. At about the same time, NASA created a Space Communi- cations and Navigation (SCaN) office within the Space Operations Mission Director- ate (SOMD) to serve as a focal point for implementing and operating this architecture. All the NASA space communications infrastructure elements have been moved to the SCaN organization and projects are being aligned with this vision.

4.1 SCaN Architecture

The SCaN architecture is a long-term vision that connects NASA spacecraft across the solar system in a system as shown in Figure 4-1.

9 Coding, Modulation, and Link Protocol Study Report

Figure 4-1: SCaN Architecture overview The dotted lines represent “long-haul” or “trunk” links between the Earth vicinity and spacecraft in deep space. The halos around targets of intense exploration (in this case, the Earth, our Moon, and Mars) represent local networks supported by additional infrastructure elements such as communication relay satellites.

While the figure shows three of these, different halos could also be deployed as needed to support new targets of interest.

Of course, there is much more to this architecture than this geometric view. This is indicated in Figure 4-2.

10 Coding, Modulation, and Link Protocol Study Report

Figure 4-2: Cross-cutting elements of the architecture

The vertical ellipses represent physical NASA infrastructure elements. Each of these provides a set of coordinated services to NASA’s space missions, enabling the missions to meet their objectives.

The Earth-based Antennas Element comprises all of NASA’s Earth-based communications and tracking assets and supporting systems. Today, NASA has a Ground Network (GN) managed by GSFC and consisting of assets that are optimized for supporting low Earth orbiting spacecraft, launches, and reentry/landings. NASA also has the Deep Space Network (DSN) which has three near-equatorial sites spaced approximately equidistance around the Earth so that at least one will be able to view a target in deep space at any time. The DSN antennas are large and optimized for long-distance communication, typically beyond geosynchronous Earth orbit.

Even today, the GN and DSN often co-support space missions. For example, the GN might support the launch of a deep space mission, turning over support to the DSN as the spacecraft leaves the Earth vicinity. In the SCaN architectural vision, all the Earth-based Antenna Element assets will be capable of such cross-support as needed.

11 Coding, Modulation, and Link Protocol Study Report

The Earth-based Relay Element consists of Geosynchronous Earth-orbiting communications relay satellites. Today, NASA has the Space Network (SN), which comprises a set of Tracking and Data Relay Satellites (TDRSs).

The Lunar Relay Satellite Element will comprise telecommunications and navigation relays in support of space missions in the lunar vicinity. NASA currently has no such assets, but will implement them when it makes sense according to its developing plans for lunar exploration.

The Mars Relay Satellite Element comprises relay satellites in orbit at Mars. Today, all NASA Mars orbiters are equipped with relay radios and, in fact, nearly all data from NASA’s Mars Exploration Rovers is returned via these orbiters. Furthermore, since the radio interfaces are defined as international (Consultative Committee for Space Data Systems Standards, or CCSDS) standards, data may also be relayed through the European Space Agency’s (ESA) Mars Express orbiter.

There are also four cross-cutting services shown in Figure 4-2.

The Networking Architecture provides the service-based interface between NASA’s missions and the underlying physical elements. This concept is only partially deployed today. This will be the glue that holds the entire system together and allows for routine interoperability between the various elements with transparency to the missions.

The Spectrum Framework defines the spectral allocations to the various links in the architecture. This is crucial to the reliable performance of the entire system. Spectral constraints ensure that all missions will have a clear signal path, unencumbered by natural or artificial interference and protected from encroachment by future users of the radio spectrum.

There is a current spectrum framework in place for NASA missions. The architectural vision requires further development of the framework to define new areas of exploration (such as the Moon) and relays.

The Security Framework will provide the necessary protection of human and robotic assets in space. This becomes more critical as humans once more venture beyond low Earth orbit.

The Navigation Architecture provides the radiometric information needed for mission operators to determine spacecraft position and trajectory. It also provides the entire system with uniform time synchronization.

12 Coding, Modulation, and Link Protocol Study Report

4.2 Need for CMLP Study

Although much has been defined in the SCaN Architecture, there is much work still to be accomplished.

One hole in particular needed to be filled before further significant investments can be made in NASA’s communications infrastructure elements. This was the definition of the physical links.

Good link designs are required for all the various links in the SCaN architecture for a number of reasons:

 Ensure adequate performance to meet anticipated mission requirements through 2030.

 Manage impacts (including cost) to the infrastructure elements.

 Manage impacts (including cost) to the space missions

 Ensure radiometrics are available as needed for navigation

 Determine required technology investments.

 Determine needs for any additional international standards.

The last item is particularly important since NASA does not expect to fulfill its ambi- tions for exploration alone.

4.3 CMLP Study Scope

Every link in the SCaN Architecture is in the scope of the CMLP Study. This includes all trunk lines and all relay (in-situ) links. The only links in the future NASA architecture that are excluded are surface-to-surface links (e.g. Earth terrestrial links or links between assets on the lunar surface that do not require a space asset for relay.) Optical links were also not considered here since it is not yet clear when they will become operational and they are still too loosely defined.

The time frame for the links in the scope of the study is from 2012 through 2030. Links that become operational before 2012 are already defined today so the study cannot influence their design. The Team chose the 2030 long-term cut-off date to

13 Coding, Modulation, and Link Protocol Study Report

both be consistent with the original SCAWG team work and to be able to include links to support humans beyond the Moon.

All reasonable coding, modulation, and multiple access schemes are within the scope of the Study. Common sense has been used to narrow the search space to a manageable size. All down-selection of such schemes was made with the concur- rence of a team of independent NASA experts.

Link protocols are described in terms of their attributes/functions, taking into consideration the particular aspects of the space environment. Discussion of the relationship between link layer and other layers is also provided. However, recommendations of specific link layer protocol standards are not given; these future recommendations will be coordinated with the SCaN Network Architecture Team (NAT) to consider related functionality in higher layer protocols.

Radiometrics and navigation are considered within the study scope. We did not consider the use of specialized navigation systems (e.g. GPS or Galileo), which will certainly be appropriate in the Earth vicinity. Only situations where the communication link is required to derive radiometric information for navigation are within the scope.

The CMLP team, to avoid duplication of prior work, was allowed to use all previous work by NASA or other experts in the areas of coding, modulation, multiple access schemes, and link protocols. The only qualification was that the team viewed the prior work as technically sound.

4.4 CMLP Study Plan

When the CMLP Study was chartered in December 2006, the plan shown in Figure 4-3 was approved. Since then, some changes have occurred, mostly due to the need to create a better interface with the international community.

All products associated with this plan are included in subsequent sections of the report.

The CMLP study progressed in three phases.

Phase 1 consisted of all the up-front work to assemble the data that would be required in the other two phases. The team developed a template for describing the links in the SCaN Architecture, including a high level description of requirements and constraints.

14 Coding, Modulation, and Link Protocol Study Report

The team identified the possible future NASA missions using the NASA Agency Mission Planning Model (AMPM) together with emerging information from the Lu- nar Architecture Team (LAT.) Requirements for future missions were gleaned from previous analyses performed at GSFC and JPL based on discussions with the teams who are developing the concepts for those missions. In the event that the AMPM lists a competitive mission opportunity, all known concepts for that opportunity (taken from science or exploration roadmaps) were considered.

The missions were grouped according to link type and application to allow for subsequent analysis in phase 2.

The team assembled a catalog of NASA’s current link types to use as a baseline.

The team also developed a set of figures of merit (FOMs) to be used in the analysis to come in phase 2.

Figure 4-3: CMLP Study schedule Concurrently with these activities, the team assembled a comprehensive catalog of codes, modulations, multiple access schemes, and link protocol attributes. These catalogs were passed to a select set of non-NASA US communications experts for review. The experts were:

 Dr. Tad White, Chief of Mathematics Research, National Security Agency (NSA)

 Dr. Roger Hammons, Communications and Networking Technology group, Johns Hopkins University (JHU) Applied Physics Laboratory (APL)

15 Coding, Modulation, and Link Protocol Study Report

 Dr. Chak Chie, LinCom Corporation

 Dr. Bob McEliece, Allen E. Puckett Professor, California Institute of Technol- ogy

 Dr. Bob Frueholz, General Manager, Communications and Networking Divi- sion, Aerospace Corporation

These experts reviewed the catalogs for completeness and for any sensitivities (e.g. national security or intellection property) of which the team was not yet aware. The experts also critiqued the FOMs. The work of the experts was invaluable for the study as the resulting catalogs are quite comprehensive.

Phase 2 consisted of the analysis of the various schemes for each link type. First, a list of “driving links” was established. These links, which represent representative links across the architecture, provide an envelope of performance and application type that allowed the team to complete its analysis in a finite time!

The work actually progressed in two stages. First, the team performed an initial down-select of schemes from the catalogs based on a common-sense application of the FOMs. For example, if two coding schemes perform equally well, but one is already an international standard, there is no need to peruse the non-standard one.

A team of internal NASA experts who were not on the CMLP team reviewed the initial down-selection. These NASA experts were

 Bernie Edwards, Goddard Space Flight Center (GSFC)

 Jason Soloff, Johnson Space Center (JSC)

 Wallace Tai, Jet Propulsion Laboratory (JPL)

 Steve Townes, Jet Propulsion Laboratory (JPL)

 Aaron Weinberg, Goddard Space Flight Center (GSFC) (Contractor)

 Dan Williams, Glenn Research Center (GRC)

After this, the team took the remaining options for each link and evaluated them with respect to the FOMs. The importance applied to each FOM was a function of the link classification (i.e. the FOM weightings the team applied were different for each link type.) This was required because different link types place differing impor-

16 Coding, Modulation, and Link Protocol Study Report tance on the attributes of the system. For example, a link from Mars to Earth will not place the same importance on coding latency as one from the Moon to the Earth since there is already much more geometric latency on the link.

This analysis resulted in the selection of a small number of recommended solutions for each link type.

Phase three consisted of writing this report and working with our international partners. To this end, Les Deutsch presented a summary of the work accomplished thus far to the 11th meeting of the Interoperability Advisory Group (IOAG) in June 2007 in Cebreros, Spain. As a result of that meeting, the team delivered a draft of this report to the IOAG and CCSDS in September 2007.

The team has now completed all planned analytical work and has completed this report. Work continues with our international partners.

4.5 CMLP Study Process

The CMLP team adopted a process similar to that used by the SCAWG in previous NASA communications and navigation studies. The process is shown graphically in Error! Reference source not found..

Figure 4-4: CMLP Process

17 Coding, Modulation, and Link Protocol Study Report

Since different links in the SCaN architecture have different requirements and special constraints, there were actually many smaller studies that needed to be accomplished. Organizing all the links into classes facilitated this. Each class contained links with similar requirements and constraints. The SCAWG process was then applied to each link class.

A set of options was developed. The options consisted of the catalogs of codes, modulations, multiple access schemes, and link protocol attributes. Other inputs to the process included the FOMs, existing information on the schemes, NASA experience with techniques, the AMPM (mission model), and the set of known regulatory issues.

The team developed evaluation scenarios for each link class. These were derived from the known or expected requirements and constraints. From these, the team developed a set of driving requirements.

The FOMs were then ranked for the specific link class. This is critical since different link classes have differing needs. For example, human missions tend to require much higher data rates than deep space robotic missions, often resulting in a need for higher spectral efficiency. Spectral efficiency would not be as important in links that only need to support deep space robotic missions.

An initial down-select was performed for each link class based on a “common- sense” application of the FOMs. This reduced the number of options to a much more manageable number.

Ultimately, a final FOM-based down-select was performed for each link class, together with supporting analysis.

NASA and external reviews were applied as mentioned above.

The final result was a small set of recommended options for each link class.

4.6 Plan for International Participation

Since international participation is a SCaN strategy, the CMLP study was set up with an eye toward this goal. Each month, a status report was delivered to the IOAG so that the international community could follow the progress of the team. The first major interaction occurred at the IOAG meeting in June 2007. There was a one-hour presentation on the study and the results obtained so far, followed by substantial discussion.

18 Coding, Modulation, and Link Protocol Study Report

International consensus on an overall plan for space communications is critical to the goals of both NASA and foreign space agencies. Space exploration has become so expensive that it is often best approached as an international endeavor. Hence there is much global interest in the CMLP study.

This IOAG meeting marked the beginning of a more regular interaction with the world’s space agencies, via future work with both the IOAG and the CCSDS – com- mencing with their comments on this report.

19 Coding, Modulation, and Link Protocol Study Report

5 NASA Links

This section provides the representative link catalogs that were used in the modulation, coding, multiple access, and link protocol down-selection processes. Sections 5.1 through 5.5 provide the Space Network (SN), Ground Network (GN), Deep Space Network (DSN), Mars in-situ, and Lunar Constellation representative link catalogs, respectively.

All links associated with the Constellation Crew Exploration Vehicle (CEV) are included in Section 5.5, the Lunar Constellation section.

5.1 NASA Space Network (SN)

Table 5-1 and Table 5-2 provide the representative SN forward and return link catalogs. These link catalogs provide a cross-section listing of future NASA missions and their respective links and do not list the entire SN mission model and link catalog.

The CEV will be supported by the SN during various mission phases, however, these links are not described in this section. Section 5.5 catalogs all CEV links including those via the SN.

5.2 NASA Ground Network (GN)

Table 5-3 and Table 5-4 provide the representative GN forward and return link catalogs. As with the SN link catalog, the GN link catalog is a cross-section representation of future GN missions and their respective links and does not list the entire GN mission model and link catalog.

5.3 NASA Deep Space Network (DSN)

Table 5-5 provides the representative Deep Space/Earth link catalog. These representative links are based upon an examination of deep space missions in the NASA Agency Mission Planning Model (AMPM).

5.4 NASA Mars Network

The Mars in-situ links and lunar in-situ links are expected to be very similar; therefore, a single representative link catalog was developed for these communications scenarios. Section 5.5 provides the lunar in-situ link catalog. This Lunar in-situ catalog was used as a representative Mars in-situ link catalog.

20 Coding, Modulation, and Link Protocol Study Report

5.5 Lunar and Constellation Links

Table 5-6 provides Lunar Relay Satellite (LRS)-to-Earth links excluding Constellation Crew Exploration Vehicle (CEV) links. If the lunar communication architecture ultimately does not include an LRS, the links in Table 5-6 are representative of the Di- rect-To-Earth (DTE) links. Table 5-7 provides LRS-lunar platform links excluding CEV links, or Direct-From-Earth (DFE) links if an LRS is used.

Table 5-8 through Table 5-10 provide CEV links as described in the Constellation Master Link Book.

Mission Planned or Support Max Data Transmit Element Receive Element Phase Rate (kbps) Data Type SN ESSP-07 LEO X 1 Ops SN ESSP-08 LEO X 1 Ops SN ESSP-09 LEO X 1 Ops SN ESSP-10 LEO X 1 Ops SN ESSP-11 LEO X 2 Ops SN ESSP-12 LEO X 2 Ops SN SYSP-01 LEO-P X 2 Ops SN SYSP-02 LEO-P X 2 Ops SN SYSP-03 LEO-P X 2 Ops SN SYSP-04 LEO-P X 2 Ops SN SYSP-05 LEO-P X 2 Ops SN GPM Core LEO X 2 Ops SN GPM CX LEO X 2 Ops SN JWST LEO X 0.25 Ops X X 32 Voice SN Shuttle LEO X X 32 Voice X X 2.4 Data SN ISS LEO X X 72 Ops Totals 13 4 5 Notes: ESSP = Earth System Science Pathfinder SYSP = Earth Systematic Project GPM = Global Precipitation Mission LEO = Low Earth Orbit LEO-P = Polar Low Earth Orbit JWST = James E. Webb Space Telescope ISS = International Space Station

Table 5-1: Representative SN Forward Link Catalog

21 Coding, Modulation, and Link Protocol Study Report

Mission Planned or Receive Support Max Data Transmit Element Element Phase Rate (Mbps) Data Type SN LEO X 300 Science ESSP-07 SN LEO X 0.064 Ops SN LEO X 300 Science ESSP-08 SN LEO X 0.064 Ops SN LEO X 600 Science ESSP-09 SN LEO X 0.064 Ops SN LEO X 600 Science ESSP-10 SN LEO X 0.128 Ops SN LEO X 1000 Science ESSP-11 SN LEO X 0.512 Ops SN LEO X 1000 Science ESSP-12 SN LEO X 1 Ops SN LEO-P X 1000 Science SYSP-01 SN LEO-P X 0.128 Ops SN LEO-P X 1000 Science SYSP-02 SN LEO-P X 0.512 Ops SN LEO-P X 1000 Science SYSP-03 SN LEO-P X 0.512 Ops SN LEO-P X 1000 Science SYSP-04 SN LEO-P X 0.512 Ops SN LEO-P X 1000 Science SYSP-05 SN LEO-P X 1 Ops GPM Core SN LEO X 0.3 Science GPM CX SN LEO X 0.3 Science GLAST SN LEO X 20 Science ISS SN LEO X 150 Science JWST SN LEO X 0.001 Ops WISE SN LEO X 120 Science General HDR SN LEO X 800** Science X X 0.032 Voice Shuttle SN LEO X X 0.032 Voice X X 0.128 Ops Totals 17 7 11 Notes: GLAST = Gamma-ray Large Area Space Telescope ISS = International Space Station WISE = Wide-field Infrared Survey Explorer HDR = Higth Data Rate ** = assumes TKUP upgrades implemented Table 5-2: Representative SN Return Link Catalog

22 Coding, Modulation, and Link Protocol Study Report

Planned Mission or Max Transmit Support Data Rate Element Receive Element Phase S-band X-band Ka-band (kbps) Data Type GN MIDEX-08 SEL1 X 4 Ops GN JWST SEL2 X 16 Ops GN MIDEX-10 SEL2 X 10 Ops GN MIDEX-12 SEL2 X 10 Ops GN SNAP (JDEM) SEL2 X 100 Ops GN ESSP-07 LEO X 1 Ops GN ESSP-08 LEO X 1 Ops GN ESSP-09 LEO X 1 Ops GN ESSP-10 LEO X 1 Ops GN ESSP-11 LEO X 2 Ops GN ESSP-12 LEO X 2 Ops GN SYSP-01 LEO X 2 Ops GN SYSP-02 LEO X 2 Ops GN SYSP-03 LEO X 2 Ops GN SYSP-04 LEO X 2 Ops GN SYSP-05 LEO X 2 Ops Totals 14 2 0 Notes: JWST = James E. Webb Space Telescope MIDEX = Medium class Explorer JDEM = Joint Dark Energy Mission SNAP = Supernova/Acceleration Probe SEL1 = Sun-Earth Lagrange Point #1 SEL2 = Sun-Earth Lagrange Point #2

Table 5-3: Representative GN Forward Link Catalog

23 Coding, Modulation, and Link Protocol Study Report

Planned or Mission Max Data Receive Support Rate Transmit Element Element Phase (Mbps) Data Type S-band X-band Ka-band MIDEX-08 GN SEL1 X X 0.213 Science GN SEL2 X 24.5 Science JWST GN SEL2 X 0.04 Ops GN SEL2 X 20 Science MIDEX-10 GN SEL2 X 0.016 Ops GN SEL2 X 30 Science MIDEX-12 GN SEL2 X 0.032 Ops GN SEL2 X 150 Science SNAP (JDEM) GN SEL2 X 0.1 Ops GN LEO X X 300 Science ESSP-07 GN LEO X 0.064 Ops GN LEO X X 300 Science ESSP-08 GN LEO X 0.064 Ops GN LEO X 600 Science ESSP-09 GN LEO X 0.064 Ops GN LEO X 600 Science ESSP-10 GN LEO X 0.128 Ops GN LEO X 1000 Science ESSP-11 GN LEO X 0.512 Ops GN LEO X 1000 Science ESSP-12 GN LEO X 1 Ops GN LEO-P X 1000 Science SYSP-01 GN LEO-P X 0.128 Ops GN LEO-P X 1000 Science SYSP-02 GN LEO-P X 0.512 Ops GN LEO-P X 1000 Science SYSP-03 GN LEO-P X 0.512 Ops GN LEO-P X 1000 Science SYSP-04 GN LEO-P X 0.512 Ops GN LEO-P X 1000 Science SYSP-05 GN LEO-P X 1 Ops Totals 14416 Table 5-4: Representative GN Return Link Catalog

24 Coding, Modulation, and Link Protocol Study Report

Range Frequency Transmit EIRP Receive Data rate Link type G/T BER Mission Type (km) (MHz) Antenna (dBW) Antenna (dB/K) (kbps)

Mars Lander 7183 70m DSN 116 Low Gain -31 0.0078125 1.00E-06 Safe Mode 400,000,000 or Orbiter 8439 Low Gain 17 70m DSN 57 0.01 1.00E-03 7183 34m DSN 110 6m 19 1000 1.00E-08 Operational 1,197,000,000 8439 6m 72 70m DSN 57 600 1.00E-08 Titan Orbiter 34316 34m DSN 122 6m 31 1000 1.00E-08 High Rate 1,197,000,000 32028 6m 90 70m DSN 62 10000 1.00E-08

7183 34m DSN 110 6m 19 6000 1.00E-08 400,000,000 8439 6m 72 70m DSN 57 6000 1.00E-08 Operational Mars Relay 7183 34m DSN 110 6m 19 20000 1.00E-08 58,000,000 Orbiter 8439 6m 72 34m DSN 51 20000 1.00E-08

34316 34m DSN 112 6m 31 10000 1.00E-08 High Rate 58,000,000 32028 6m 90 34m DSN 56 500000 1.00E-08 Table 5-5: Representative DSN Link Catalog

Low Rate High Rate Transmission Source Number of Data Rate Number of Data Rate Sources (Mbps) Sources (Mbps) LSAM 1 0.592 1 22 EVA suit 4 0.002 Rover 2 1.75 2 20.5 Surface Mobility Carrier 2 0.25 2 4.5 O2 Excavator 1 0.25 1 9.5 O2 Mobile Servicer 1 0.25 1 11 H2/H2O Excavator 1 0.25 1 9.5 H2/H2O Extraction Unit 1 0.25 1 4.5 H2/H2O Mobile Servicer 1 0.25 1 12.5 Habitat 1 2.824 1 135 TOTAL 15 10.799 11 229 Ranging links 13 Table 5-6: Lunar Platform-to-LRS or DTE Links

25 Coding, Modulation, and Link Protocol Study Report

Low Rate High Rate Destination Data Rate Data Rate Number Number (Mbps) (Mbps) LSAM 1 1.020 1 28 EVA suit 4 0.010 Rover 2 1.558 2 25.5 Surface Mobility Carrier 2 0.4 O2 Excavator 1 0.2 O2 Mobile Servicer 1 0.2 H2/H2O Excavator 1 0.2 H2/H2O Extraction Unit 1 0.2 H2/H2O Mobile Servicer 1 0.2 Habitat 1 2.949 1 100 TOTAL 15 8.725 4 153.5 Ranging links 13 Table 5-7: LRS-to-Lunar Platform or DFE Links

Frequency3 Transmit Receive Mode/Rate Mission Phase Link Name Link type (MHz) Antenna Antenna Mode (kbps) Low Earth Orbit TDRSS Links SN-CEV S-Band Contingency Forward Contingency 2106.4 TDRS-SA Low Gain LDR 18 CEV-SN S-Band Contingency Return Contingency 2287.5 Low Gain TDRS-SA LDR 24 SN-CEV S-Band Operational P2P Forward LDR Operational 2106.4 TDRS-SA Low Gain LDR 72 CEV-SN S-Band Operational P2P Return LDR Operational 2287.5 Low Gain TDRS-SA LDR 192 SN-CEV S-Band Operational P2P Forward HDR Operational 2106.4 TDRS-SA High Gain HDR 1,000 CEV-SN S-Band Operational P2P Return HDR Operational 2287.5 High Gain TDRS-SA HDR 1,000

SN-CEV Ka-Band High Rate P2P Forward High Rate 23205.0 TDRS-SA 0.75m Dish HDR 6,000 CEV-SN Ka-Band High Rate P2P Return High Rate 25600.0 0.75m Dish TDRS-SA HDR 25,000

GN Ground Links GN-CEV S-Band Contingency Uplink Emergency 2106.4 11.3m Dish Low Gain LDR 18 CEV-GN S-Band Contingency Downlink Emergency 2287.5 Low Gain 11.3m Dish LDR 24 GN-CEV S-Band Operational P2P Uplink LDR Emergency 2106.4 11.3m Dish Low Gain LDR 72 CEV-GN S-Band Operational P2P Downlink LDR Emergency 2287.5 Low Gain 11.3m Dish LDR 192 GN-CEV S-Band Operational P2P Uplink HDR Emergency 2106.4 11.3m Dish High Gain HDR 1,000 CEV-GN S-Band Operational P2P Downlink HDR Emergency 2287.5 High Gain 11.3m Dish HDR 1,000

Rendezvous Links ISS-CEV S-Band Operational P2P Forward VLDR Rendezvous 2030.4 Low Gain Low Gain LDR 24 CEV-ISS S-Band Operational P2P Return VLDR Rendezvous 2205.0 Low Gain Low Gain LDR 24 ISS-CEV S-Band Operational P2P Forward LDR Rendezvous 2030.4 Low Gain Low Gain HDR 192 CEV-ISS S-Band Operational P2P Return LDR Rendezvous 2205.0 Low Gain Low Gain HDR 192 ISS-CEV S-Band Operational P2P Forward HDR Rendezvous 2030.4 Low Gain Low Gain Video 6,000 CEV-ISS S-Band Operational P2P Return HDR Rendezvous 2205.0 Low Gain Low Gain Video 6,000

LSAM-CEV S-Band Operational P2P Forward VLDR Rendezvous 2030.4 Low Gain Low Gain LDR 24 CEV-LSAM S-Band Operational P2P Return VLDR Rendezvous 2205.0 Low Gain Low Gain LDR 24 LSAM-CEV S-Band Operational P2P Forward LDR Rendezvous 2030.4 Low Gain Low Gain HDR 192 CEV-LSAM S-Band Operational P2P Return LDR Rendezvous 2205.0 Low Gain Low Gain HDR 192 LSAM-CEV S-Band Operational P2P Forward HDR Rendezvous 2030.4 Low Gain Low Gain Video 6,000 CEV-LSAM S-Band Operational P2P Return HDR Rendezvous 2205.0 Low Gain Low Gain Video 6,000

Table 5-8: CEV LEO Mission Phase Link Catalog

26 Coding, Modulation, and Link Protocol Study Report

Frequency3 Transmit Receive Mode/Rate Mission Phase Link Name Link type (MHz) Antenna Antenna Mode (kbps) Trans Lunar/Lunar Orbit CTN Ground Links CTN-CEV S-Band Contingency Uplink Contingency 2106.4 18.0m Dish Low Gain LDR 18 CEV-CTN S-Band Contingency Downlink Contingency 2287.5 Low Gain 18.0m Dish LDR 24 CTN-CEV S-Band Operational P2P Uplink LDR Operational 2106.4 18.0m Dish Low Gain LDR 72.0 CEV-CTN S-Band Operational P2P Downlink LDR Operational 2287.5 Low Gain 18.0m Dish LDR 192.0 CTN-CEV S-Band Operational P2P Uplink HDR Operational 2106.4 18.0m Dish High Gain HDR 1,000 CEV-CTN S-Band Operational P2P Downlink HDR Operational 2287.5 High Gain 18.0m Dish HDR 1,000

CTN-CEV Ka-Band High Rate P2P Uplink High Rate 23205.0 TDRS-SA 0.75m Dish HDR 6,000 CEV-CTN Ka-Band High Rate P2P Downlink High Rate 25600.0 0.75m Dish TDRS-SA HDR 25,000

CTN-LSAM Ground links CTN-LSAM S-band Contingency Uplink Contingency 2106.4 18.0m Dish Low Gain LDR 18 LSAM-CTN S-Band Contingency Downlink Contingency 2287.5 Low Gain 18.0m Dish LDR 24 CTN-LSAM S-Band Operational P2P Uplink LDR Operational 2106.4 18.0m Dish Low Gain LDR 72.0 LSAM-CTN S-Band Operational P2P Downlink LDR Operational 2287.5 Low Gain 18.0m Dish LDR 192.0 CTN-LSAM S-band Operational P2P Uplink HDR Operational 2106.4 18.0m Dish High Gain HDR 1,000 LSAM-CTN S-Band Operational P2P Downlink HDR Operational 2287.5 0.75m Dish 18.0m Dish HDR 1,000

CTN-LSAM Ka-Band High Rate P2P Uplink High Rate 23205.0 18.0m Dish 0.75m Dish HDR 6,000 LSAM-CTN Ka-Band High Rate P2P Downlink High Rate 25600.0 0.75m Dish 18.0m Dish HDR 25,000

DSN Ground Links DSN-CEV S-Band Contingency Uplink Emergency 2106.4 34.0m Dish Low Gain LDR 18 CEV-DSN S-Band Contingency Downlink Emergency 2287.5 Low Gain 34.0m Dish LDR 24 DSN-CEV S-Band Operational P2P Uplink LDR Emergency 2106.4 34.0m Dish Low Gain LDR 72 CEV-DSN S-Band Operational P2P Downlink LDR Emergency 2287.5 Low Gain 34.0m Dish LDR 192 DSN-CEV S-Band Operational P2P Uplink HDR Emergency 2106.4 34.0m Dish High Gain HDR 1,000 CEV-DSN S-Band Operational P2P Downlink HDR Emergency 2287.5 High Gain 34.0m Dish HDR 1,000

GN Ground Links GN-CEV S-Band Operational P2P Uplink HDR Emergency 2106.4 11.3m Dish High Gain HDR 1000.0 CEV-GN S-Band Operational P2P Downlink HDR Emergency 2287.5 High Gain 11.3m Dish HDR 1000.0

Lunar Relay Links LCNS-CEV S-Band Contingency Forward Contingency 2106.4 12.0m Dish Low Gain LDR 18.0 CEV-LCNS S-Band Contingency Return Contingency 2287.5 Low Gain 12.0m Dish LDR 24.0 LCNS-CEV S-Band Operational P2P Forward LDR Operational 2106.4 12.0m Dish Low Gain LDR 72.0 CEV-LCNS S-Band Operational P2P Return LDR Operational 2287.5 Low Gain 12.0m Dish LDR 192.0 LCNS-CEV S-Band Operational P2P Forward HDR Operational 2106.4 12.0m Dish High Gain HDR 1000.0 CEV-LCNS S-Band Operational P2P Return HDR Operational 2287.5 0.75m Dish 12.0m Dish HDR 192.0

LCNS-CEV Ka-Band High Rate P2P Forward High Rate 23205.0 12.0m Dish 0.75m Dish HDR 6000.0 CEV-LCNS Ka-Band High Rate P2P Return High Rate 25600.0 0.75m Dish 12.0m Dish HDR 25000.0

Rendezvous Links LSAM-CEV S-Band Operational P2P Forward VLDR Rendezvous 2030.4 Low Gain Low Gain LDR 24 CEV-LSAM S-Band Operational P2P Return VLDR Rendezvous 2205.0 Low Gain Low Gain LDR 24 LSAM-CEV S-Band Operational P2P Forward LDR Rendezvous 2030.4 Low Gain Low Gain HDR 192 CEV-LSAM S-Band Operational P2P Return LDR Rendezvous 2205.0 Low Gain Low Gain HDR 192 LSAM-CEV S-Band Operational P2P Forward HDR Rendezvous 2030.4 Low Gain Low Gain Video 6,000 CEV-LSAM S-Band Operational P2P Return HDR Rendezvous 2205.0 Low Gain Low Gain Video 6,000

Table 5-9: CEV Trans Lunar/Lunar Orbit Mission Phase Link Catalog

27 Coding, Modulation, and Link Protocol Study Report

Frequency3 Transmit Receive Mode/Rate Mission Phase Link Name Link type (MHz) Antenna Antenna Mode (kbps) Launch/Ascent TDRSS Links SN-CEV S-Band Contingency Forward Contingency 2106.4 TDRS-SA Low Gain LDR 18 CEV-SN S-Band Contingency Return Contingency 2287.5 Low Gain TDRS-SA LDR 24 SN-CEV S-Band Operational P2P Forward LDR Operational 2106.4 TDRS-SA Low Gain LDR 72 CEV-SN S-Band Operational P2P Return LDR Operational 2287.5 Low Gain TDRS-SA LDR 192

GN Ground/ROCC CLV-GN S-Band Operational P2P Downlink LDR Operational 2215.0 Low Gain 9.0m Dish LDR 192 CLV-GN S-Band Operational P2P Downlink HDR Operational 2265.0 Low Gain 9.0m Dish HDR 20,000 GN-CLV UHF FTS Command Uplink Range Safety TBD Dish FTS Omni LDR N/A

Earth Entry TDRSS Links SN-CEV S-Band Contingency Forward Contingency 2106.4 TDRS-SA Low Gain LDR 18 CEV-SN S-Band Contingency Return Contingency 2287.5 Low Gain TDRS-SA LDR 24 SN-CEV S-Band Operational P2P Forward LDR Operational 2106.4 TDRS-SA Low Gain LDR 72 CEV-SN S-Band Operational P2P Return LDR Operational 2287.5 Low Gain TDRS-SA LDR 192

Orion to NEO 34m DSN Ground Links DSN-CEV S-Band Contingency Uplink Nominal 2106.4 34.0m Dish Low Gain LDR 18 CEV-DSN S-Band Contingency Downlink Nominal 2287.5 Low Gain 34.0m Dish LDR 24 DSN-CEV S-Band Operational P2P Uplink LDR Nominal 2106.4 34.0m Dish Low Gain LDR 72 CEV-DSN S-Band Operational P2P Downlink LDR Nominal 2287.5 Low Gain 34.0m Dish LDR 192 DSN-CEV S-Band Operational P2P Uplink HDR Nominal 2106.4 34.0m Dish High Gain HDR 1,000 CEV-DSN S-Band Operational P2P Downlink HDR Nominal 2287.5 High Gain 34.0m Dish HDR 1,000

34m DSN-CEV Ka-Band High Rate P2P Uplink High Rate 23205.0 34m Dish High Gain HDR 6,000 CEV-34m DSN Ka-Band High Rate P2P Downlink High Rate 25600.0 High Gain 34m Dish HDR 25,000

70m DSN Ground Links 70m DSN-CEV S-Band Contingency Uplink Nominal 2106.4 70.0m Dish Low Gain LDR 18 CEV-70m DSN S-Band Contingency Downlink Nominal 2287.5 Low Gain 70.0m Dish LDR 24 70m DSN-CEV S-Band Operational P2P Uplink LDR Nominal 2106.4 70.0m Dish Low Gain LDR 72 CEV-70m DSN S-Band Operational P2P Downlink LDR Nominal 2287.5 Low Gain 70.0m Dish LDR 192 70m DSN-CEV S-Band Operational P2P Uplink HDR Nominal 2106.4 70.0m Dish High Gain HDR 1,000 CEV-70m DSN S-Band Operational P2P Downlink HDR Nominal 2287.5 High Gain 70.0m Dish HDR 1,000

70m DSN-CEV Ka-Band High Rate P2P Uplink Nominal 23205.0 70.0m Dish High Gain HDR 6,000 CEV-70m DSN Ka-Band High Rate P2P Downlink Nominal 25600.0 High Gain 70.0m Dish HDR 25,000

Table 5-10: Other CEV Links

5.6 Multiple Access Regimes

This section defines the communication network evaluation regimes that were used in the CMLP study, particularly for the purpose of evaluating multiple access techniques.

5.6.1 Near Earth Relay (NER) Space Network (SN) Three equally-spaced near Earth relay satellites at geosynchronous altitude shall be capable of supporting spacecraft in LEO continuously (24/7). The NER spacecraft provide tracking services to user spacecraft as well as command and telemetry (TT&C) services at S-band and high data rate forward and return mission data services at Ku-band and Ka-band (23/26GHz). Multiple access services shall be provided using an S-band phased array antenna and a multi-band reflector antenna(s) on the NER at S-band, Ku-band, and Ka-band.

28 Coding, Modulation, and Link Protocol Study Report

5.6.2 Near Earth Ground Network The GN consists of 50 ground station antennas and 30 unique antenna systems distributed at seven geographic antenna locations that provide ground-based space communications for NASA missions, including LEO and GEO orbital missions, sub- orbital missions, and launch support. These stations provide TT&C and range support in several frequency bands: X-band, S-band, L-band, VHF, UHF, and C-band.

5.6.3 Lunar DTE/DFE Three equally-spaced lunar DTE/DFE ground antennas shall be capable of supporting lunar space vehicles in transit to/from the moon or in Low Lunar Orbit (LLO) or on the lunar surface. The SCaN Architecture requires lunar DTE/DFE ground antennas to provide tracking services to the user spacecraft as well as command and telemetry (TT&C) services at S-band and high data rate forward and return mission data services at Ka-band (23/26GHz). Multiple access services will be provided through a DTE/DFE ground station at S-band or Ka-band.

5.6.4 Lunar Relay A system of Lunar Relay Satellites (LRSs) will relay communications to and from lunar space vehicles near the moon or in Low Lunar Orbit (LLO) or on the lunar surface. LRSs will provide tracking services to user spacecraft as well as command and telemetry (TT&C) services at S-band and high data rate forward and return mission data services at Ka-band (23/26GHz), providing the lunar user community with a communications link to Earth. Multiple access services will be provided through the LRS satellites at S-band and Ka-band. A separate Ka-band trunk line (37/40 GHz) will be used to relay the data to and from Earth

5.6.5 Deep Space DTE/DFE Ground antennas at Deep Space Network (DSN) complexes will support Mars space vehicles in transit to/from planetary bodies or in planetary orbit or on the planet’s surface. DSN ground antennas will provide DTE/DFE tracking services to user spacecraft as well as command and telemetry (TT&C) services at X-band (7/8 GHz) and high data rate forward and return mission data services at Ka-band (31/34 GHz). Multiple access services will be provided using DSN apertures at S-band and Ka-band.

At this time, deep space multiple access is expected to be needed only for Mars missions, though multiple access techniques developed for Mars missions are likely to be applicable elsewhere should a need arise.

29 Coding, Modulation, and Link Protocol Study Report

5.6.6 Mars Relay A constellation of science orbiters with relay payloads and one or more Mars Areo- stationary Relay Satellites, or MARSats will serve as communication relays between Mars exploration elements (landers, rovers, balloons, airplanes, etc.) and the Earth and will track exploration elements.

30 Coding, Modulation, and Link Protocol Study Report

6 Current CMLP Schemes

6.1 Space Network (SN) Schemes

The NASA SN provides continuous global communications and tracking services to LEO satellites through the use of nine on-orbit Tracking and Data Relay Satellites and three backhaul ground stations located at White Sands, New Mexico (site of two SN ground stations) and Guam, Marianas Islands. Error! Reference source not found. provides an overview of the NASA SN.

Figure 6-1: NASA SN Overview Figure 6-2 provides an illustration of a second generation Tracking and Data Relay Satellite (TDRS). The spacecraft provides S-band Multiple Access (MA) forward and return services via a phased array antenna on the body of the spacecraft and S-band, Ku-band and Ka-band forward and return services via two steerable tri-band single access antennas mounted on booms emerging from the east and west sides of the spacecraft body. A 2m Ku-band Space-to-Ground Link (SGL) antenna is used for transmission to and reception from the SN ground terminal.

The current SN provides communications and tracking services using the modulation, coding, multiple access and link protocol techniques as described in Table 6-1. The CMLP techniques currently supported by the SN were selected based upon many factors; the remainder of this section will describe these factors.

31 Coding, Modulation, and Link Protocol Study Report

The primary modulation techniques supported by the SN are filtered BPSK, QPSK and OQPSK. These modulation techniques offer a superior balance of power efficiency, spectral efficiency and hardware complexity for NASA’s current and future missions. Additionally, a heritage has been established through the successful and reliable use of these techniques within the SN, particularly with manned platforms.

980105-328

WEST SA ANTENNA FORWARD OMNI ANTENNA

MA RETURN ANTENNA

MA FORWARD ANTENNA SA COMPARTMENT (SAC) SOUTH SOLAR WING

SGL ANTENNA

EAST SA ANTENNA SPACECRAFT AFT BUS OMNI (hidden)

NORTH SOLAR WING (ORBIT DIRECTION) Figure 6-2: Second Generation TDRS (courtesy Boeing Satellite Systems)

While standard filtered BPSK, QPSK and OQPSK have many advantages over other modulation types, they have a significant disadvantage in that they are non-constant envelop modulation techniques. This non-constant envelope characteristic translates into power or spectrum inefficiency relative to constant envelope modulation techniques. Traditionally, this power or spectrum inefficiency was accepted over the increased hardware complexity of constant envelope modulation techniques, however, this trade is no longer as clear as it once was.

The SN currently also supports GN-style residual carrier modulation techniques in both forward and return directions. These modulations are supported by the SN to enable an SN global commanding and telemetering capability to GN missions. While residual carrier modulation techniques are certainly not the most power or spectral efficient modulation techniques, SN support of these techniques introduces a diversity and reliability during spacecraft emergencies that outweighs the power and spectral inefficiencies.

NASA continuously works to expand and improve the customer services offered by the SN. As part of this effort, NASA initiated the TDRSS K-band Upgrade Project Augmentation (TKUP-A) in 2005. One objective of this project is to replace the aging

32 Coding, Modulation, and Link Protocol Study Report

SN high data rate receivers with more capable receivers that will support all current SN modulation techniques and will also support 8PSK. The rationale for supporting 8PSK is to enable increased customer service data rates over the TDRS Ku/Ka-band 225 MHz and Ka-band 650 MHz channels. Support of the 8PSK modulation technique is currently scheduled to begin in 2012/2013 timeframe.

33 Coding, Modulation, and Link Protocol Study Report

Network Supported Techniques Comments Parameter With sufficient transmitter filtering to meet NTIA emission BPSK mask requirements With sufficient transmitter filtering to meet NTIA emission QPSK/OQPSK mask requirements PCM/PSK/PM Modulation Supported only in forward and return direction PCM/PM SN is currently specifying requirements for future SN hardware, which will support 8PSK. This is expected to be 8PSK available shortly after Spring 2012. To be supported at Ku- band and Ka-band only Uncoded Rate 1/2 Convolutional Standard constraint length = 7 code Rate 1/3 Convolutional Supported on S-band links only Concatenated R-S + Supported for select service modes Coding Conv Rate 7/8 TPC SN is currently specifying requirements for future SN Rate 7/8 LDPCC hardware, which will support high performance codes. SN S-band LDPCC service modes are expected to be available Rate ½ LDPCC by Spring 2012. SN Ku/Ka-band TPC and LDPCC are expected to be available shortly after Spring 2012. Scheduled Single Access antennas serve multiple users Time Shared through time sharing A form of SDMA is used by the Multiple Access antenna which can form 5 simultaneous return beams through a SDMA second generation TDRS and a nearly unlimited amount of Multiple return beams through a first generation TDRS Access PN codes used for MA return services to further discrimi- CDMA nate between antenna beams On both forward and return, user services, command, and FDMA telemetry are separated by frequency for simultaneous transmission via TDRS FDMA/CDMA CDMA is supported at S-band, Ku-band, and Ka-band AOS The SN provides a bitstream service that is transparent to TC the link protocol; any may be used. However, those listed Link Protocol are typical. AOS = Advanced Orbiting Systems CCSDS TM 732.0-B-2, TC = Telecommand CCSDS 232.0-B-1, TM = Te- lemetry Space Data Link Protocol CCSDS 132.0-B-1 Table 6-1: Current SN-Supported CMLP Techniques . Another aspect of the TKUP-A project is the expansion of SN supported customer service coding options. The SN currently supports traditional space communication coding techniques including rate ½ convolutional coding, Reed-Solomon coding and concatenated coding (Reed-Solomon with convolutional). While these coding techniques offer a good combination of coding gain and hardware complexity and have

34 Coding, Modulation, and Link Protocol Study Report

a tremendous legacy of successful use in space communications systems, new high performance codes are rapidly emerging which can significantly reduce the customer power burden with little to no increase in customer hardware complexity. It must be noted, however, that the TKUP-A project is still in the requirements definition and concept demonstration phase and funding to implement the TKUP-A capabilities is not assured.

Early in the TKUP-A project, a detailed evaluation of current and emerging coding techniques was performed to determine which codes offered the largest benefits with the fewest drawbacks. In particular, the TKUP-A project pursued signal structures that reduced customer burden (power and hardware complexity) and / or enabled increased data rates over those currently supported by the system. The NASA TKUP-A team ultimately selected three codes for implementation as part of the upgrade project. These three codes are rate 7/8 TPC, rate 7/8 LDPCC and rate ½ LDPCC. Pending the outcome of a January 2008 demonstration, implementation of one of the rate 7/8 codes may be eliminated.

Customers can access SN resources in a variety of ways as identified in Table 6-1. Most notably, each TDRS is equipped with a forward and return service multiple access phased array antenna. The phased array antenna enables support to multiple customer platforms through the use of beamforming to introduce spatial discrimination. Additionally, all customers using the TDRS multiple access antenna must be PN spread to further introduce discrimination among customers and to ensure PFD limits are met.

6.2 Ground Network (GN) Schemes

The NASA GN provides communications and tracking services to a diverse set of national, international and commercial missions during various phases of their missions at various frequency bands. Figure 6-3 provides an overview of the geographic distribution of the NASA GN terminals.

35 Coding, Modulation, and Link Protocol Study Report

Figure 6-3: Geographic Distribution of NASA GN Terminals

The current GN provides communications and tracking services using the modulation, coding, multiple access and link protocol techniques as described in Table 6-2. The CMLP techniques currently supported by the GN were selected based upon many factors; the remainder of this section will discuss the rationale for the current GN-supported CMLP techniques.

36 Coding, Modulation, and Link Protocol Study Report

Network Supported Techniques Comments Parameter BPSK QPSK/OQPSK Modulation techniques are available selectively depending PCM/PSK/PM Modulation on ground location, frequency band, and direction of com- PCM/PM munications (command or telemetry) FM AM Rate ½ Convolutional Reed-Solomon Coding Concatenated R-S + Not available for some mobile ground terminals Conv Time Shared Ground antennas schedule service to multiple users A form of SDMA is used through the use of several antennas Multiple Ac- SDMA at a single ground location cess In some cases, a single antenna will support more than one FDMA frequency band AOS The GN provides a bitstream service that is transparent to Link Protocol TC the link protocol; any may be used. However, those listed TM are typical. Table 6-2: Current GN-Supported CMLP Techniques

6.3 Deep Space Network (DSN) Schemes

The NASA Deep Space Network (DSN) is an international network of antennas that supports interplanetary spacecraft missions and select Earth-orbiting missions. The DSN utilizes three deep-space communications facilities placed approximately 120 degrees apart around the world at Goldstone, CA; Madrid, Spain; and Canberra, Australia. Figure 6-4 illustrates the geographic distribution of DSN sites.

Figure 6-4: Geographic Distribution of NASA DSN Terminals

37 Coding, Modulation, and Link Protocol Study Report

The current DSN provides communications and tracking services using the modulation, coding, multiple access and link protocol techniques as described in Table 6-3. Network Supported Techniques Comments Parameter PCM/PSK/PM Sine wave or square wave subcarrier. PCM/PM Modulation QPSK OQPSK Convolutional Rate 1/2 to 1/6, constraint length up to 15 Reed-Solomon Concat R-S + Conv Coding Turbo Codes Rate 1/2 to 1/6 A ground antenna could conceivably communicate with Time Shared multiple missions on a timesharing basis Each ground complex consists of multiple stations SDMA equipped with antennas and receiving systems Multiple Ac- Depending on the antenna, S-band, X-band, and K-band cess FDMA command and telemetry can be supported PCM/PSK/PM Sine wave or square wave subcarrier. AOS Frame services indicated. Bit Stream service is slated for Link Protocol TC decommissioning and is not available to new customers; TM link protocol is transparent in that case. Table 6-3: Current DSN-Supported CMLP Techniques Reference: DSMS Services and Capabilities, Edward B. Luers, May 2006

6.4 Mars Network

The current Mars network provides proximity link communication services using the modulation, coding, multiple access and link protocol techniques as described in Table 6-4.

Network Supported Techniques Comments Parameter BPSK Residual or suppressed carrier Modulation QPSK Coding Convolutional Rate ½, constraint length 7 Multiple Ac- Time shared cess Proximity-1 Proximity-1 = Proximity-1 Space Link Protocol – Data AOS Link Protocol Link Layer CCSDS 211.0-B-3. AOS, TC and TM are TC provided with DSN. TM Table 6-4: Current Mars Network Proximity Link CMLP Techniques

38 Coding, Modulation, and Link Protocol Study Report

7 CMLP Schemes

7.1 Modulation

Table 7-1 describes the candidate modulation techniques considered in the CMLP study. The CMLP modulation catalog contains 73 specific modulations from 12 modulations groups. For this study, we separately treated modulations received by different structures, despite the fact that the transmitted modulation may be identical. This is because an optimal receiver for a modulation (one heavily leaden with memory, for example) might perform well but be too complex to realize in practice. For such cases, we separately evaluated an optimal trellis receiver and an integrate and dump receiver.

There are several ways to group the modulations – for example by modulation type, by modulation order, by spectral efficiency, or by standardization status. In Table 7-1, the modulations are grouped by modulation order, which is the number of bits per modulation symbol, since historically there has been a close association between the modulation order and the space application. For example, deep space communication has been accomplished almost exclusively with modulation orders of 2 or 4.

In describing the modulations, however, it is easier to group them by their type, i.e., the family of phase shift keying (PSK) modulations, the QAM modulations, etc. Of the modulations in Table 7-1 the following have been recommended by the Consul- tative Committee for Space Data Systems (CCSDS): PCM/PSK/PM, PCM/PM, BPSK, OQPSK, GMSK, and 8-PSK.

QPSK and 16-QAM have not been recommended in CCSDS standards but are in non-CCSDS standards. OQPSK (with pulse shaping) instead of QPSK is recommended by CCSDS to reduce sidelobe regeneration after nonlinear amplifiers.

GMSK is described in detail in Appendix 7, along with a way to integrate ranging with a GMSK data signal.

7.1.1 PSK Phase Shift Keying (PSK) is a digital phase modulation technique in which the signal waveforms are defined as follows:

39 Coding, Modulation, and Link Protocol Study Report

 m  )1(2  2cos[)()(  tftgtS  ] , 0 ≤ t ≤ T m c M g(t) is the pulse shape of the transmitted symbol. M is the modulation order. T is the symbol duration.

In the simplest case, g(t) is a rectangular pulse. As can be seen from the equation above, in that case the signal rapidly ranges from -1 to +1, which results in a constant envelope. A constant envelope is a beneficial property because it allows the power amplifiers to operate saturated without introducing distortion. The amplification of a non-constant envelope signal results in AM/AM and AM/PM distortion if the amplifier is operated at saturation. Possible remedies in that situation are to (1) back off the power to the linear region of the amplifier, (2) accept the losses due to nonlinear distortion, or (3) pre-distort the signal to account for the subsequent nonlinearity of amplification.

A popular modulation for deep space communication is binary PSK (BPSK). BPSK is PSK with M=2. We say that the modulation order is 2, corresponding to the two phases (0 and Pi). A single input bit determines which phase is transmitted, corresponding to m=1 and m=2 in the equation above.

PSK modulation with M=4 is quaternary PSK (QPSK). In this case, four phases are allowed, and two bits are required to determine the symbol transmitted during a symbol epoch. This corresponds to m=1, 2, 3, 4 in the equation above. Alternatively, the equation above may be rewritten into quadrature cosine and sine components, also called in-phase and quadrature components, using simple trigonometric expansion. The in-phase and quadrature components are then each controlled by a single bit.

When QPSK is used, the transmitted phases conventionally are defined as Pi/4 3Pi/4, 5Pi/4, and 7Pi/4. This does not agree with the equation above as written, because the phases just mentioned are an additional Pi/4 larger than the ones the equation above contemplates. This is of no consequence to the transmission or reception of PSK signals, however, because an additional arbitrary phase must be recov- ered at the receiver anyway. In general, the M-ary PSK phase corresponding to m=1 in the equation above is not 0, but rather Pi/M.

When QPSK is used, transitions from any of the four phases to any of the other four phases is possible, depending on the modulating bit pattern. A transition of Pi, however, can cause distortion in the amplifier, due to the swing through the origin. This temporarily destroys the constant envelope property of PSK. For this reason, the in-phase and quadrature components may be offset from each other by a half- symbol (T/2). This is called offset QPSK (OQPSK) in the academic literature, and

40 Coding, Modulation, and Link Protocol Study Report

staggered QPSK (SQPSK) in SN documentation. The staggering prevents symbol phase transitions of Pi, and limits them to transitions of Pi/2, which do not go through the origin. In this way, the constant envelope property can be maintained.

A principal problem with PSK is that it is not spectrally efficient. BPSK, for example, achieves only 0.05 bits/sec/Hz using the 99% BW definition of bandwidth usage. A Square-Root-Raised-Cosine (SRRC) shaping/filtering substantially improves the spectral efficiency, to 0.79 bits/sec/Hz for a rolloff factor of 0.5 and to 0.89 bits/sec/Hz for a rolloff factor of 0.2. With perfect timing recovery and use of a matched filter, no loss is seen with the SRRC.

Filtered PSK (BPSK,QPSK, OQPSK) is 1.) moderately spectrally efficient; 2.) moderately susceptible to phase noise and nonlinear distortions; 3.) very power efficient; 4.) technologically mature; 5.) easy to implement; 6.) compatible with existing SN, GN and DSN infrastructure.

Various forms of 4-ary PSK are OQPSK, baseband filtered OQPSK with linear phase modulator (OQPSK/PM), SOQPSK-A, SOQPSK-B, FQPSK-B. These forms of PSK are 1.) highly bandwidth efficient; 2.) constant envelope; 3.) less susceptible to nonlinear distortions; 4.) very power efficient if trellis receiver is used. OQPSK/PM is technologically mature. SOQPSK-A, SOQPSK-B and FQPSK-B are less technologically mature and can be difficult to implement at high rates.

Differential PSK (DPSK) is a form of PSK in which the data does not directly modu- late the phase, but is first sent through a differential encoder (an accumulator) and then to the phase modulation. DPSK 1.) can be spectrally efficient; 2.) is susceptible to phase noise and nonlinear distortions; 3.) is relatively power inefficient; 4.) typically used when low-complexity, high-reliability link design is required.

Higher order PSK modulation; such as filtered 8PSK, 16PSK are 1.) highly spectrally efficient in a linear power amplifier operational mode; 2.) highly susceptible to phase noise and nonlinear distortions; 3.) less power efficient; 4.) less technologically mature in space applications; 5.) typically used for high data rate users with a relatively small spectrum allocation.

Specially filtered QPSK modulations also exist. Principal among these is the Feher- patented QPSK, or FQPSK. This modulation is highly spectrally efficient.

7.1.2 PCM/PSK/PM PCM/PSK/PM is defined as NRZ data that is PSK modulated on a squarewave or sinewave subcarrier which is then phase modulated on the RF carrier. Its cryptic name refers to a pulse code modulated (PCM) signal, i.e., one that is a digital signal

41 Coding, Modulation, and Link Protocol Study Report

as opposed to analog, in which the data are phase shift keyed (PSK) onto a subcarrier. The subcarrier is then phase modulated (PM) onto the carrier. To specify whether the subcarrier modulation is squarewave or sinewave, another term is sometimes added so that the name becomes PCM/PSK/PM-sinewave or PCM/PSK/PM-squarewave.

If there is no subcarrier. i.e., data are directly modulated on the carrier but still using a residual carrier, the name becomes PCM/PM. To differentiate between bi-phase and NRZ, an extra term is added so that the name becomes PCM/PM/NRZ or PCM/PM/Bi-phase. If there is no residual carrier, then it simply becomes BPSK.

PCM/PSK/PM signal waveforms can be described as:

 2)( [sin(ctAtS ) cos(mp  )()( cos(cttdt )sin(mp(t)]

A is the amplitude. m is the modulation index in radian. d(t) is the NRZ binary valued data sequence with symbol period T. P(t) is the subcarrier waveform(squarewave or sinewave).

PCM/PSK/PM is the most popular modulation used in deep space communication. Some of its features are: 1.) less bandwidth efficient as compared to traditional PSK modulations; 2.) less power efficient compared to suppressed carrier modulations; 3.) highly susceptible to nonlinear distortions; 4.) technologically mature; 5.) compatible with existing SN, GN and DSN infrastructure.

PCM/PSK/PM and PCM/PM/NRZ (next subsection) are both residual carrier modulations. They will continue to be used for very low rate emergency links, in which a residual carrier is needed to adequately reconstruct the carrier for coherent demodulation.ii This need is expected to continue throughout the time frame considered for this study, though its need depends on the specific nature of the spacecraft and ground oscillators, channel impairments (e.g., solar scintillation), and data rate.

7.1.3 PCM/PM/NRZ or Bi-phase

PCM/PM/NRZ or Bi-phase is defined as NRZ or Bi-phase data which is phase modulated directly on a residual RF carrier. PCM/PM/NRZ or Bi-phase signal waveforms can be described as:

 2)( [sin(c ) cos(  )() cos(c )sin(mttdmtAtS )]

42 Coding, Modulation, and Link Protocol Study Report

A is the amplitude. m is the modulation index in radian. d(t) is formatted in the form of NRZ or Bi-phase.

PCM/PM/NRZ or Bi-phase is 1.) less bandwidth efficient as compared to traditional PSK modulations; 2.) less power efficient compared to suppressed carrier modulations; 3.) highly susceptible to nonlinear distortions; 4.) technologically mature; 5.) compatible with existing SN, GN and DSN infrastructure.

43 CMLP Final Report

Table 7-1: CMLP Modulation Catalog

CMLP Study Page 44 August 2007 CMLP Final Report

7.1.4 FSK and GMSK

Frequency Shift Keying (FSK) is the binary form of Frequency Modulation. FSK signal waveform can be described as:

2   )(  cos[    ])(22    

Where, Eb is the bit energy. Tb is the bit duration. k f is the modulation index.

A type of FSK is sometimes used for communications during the entry-descent- landing portion of deep space missions. In that scenario, simple tones or combinations of tones are transmitted to indicate the success or failure of particular events such as the opening of a parachute or firing of a thruster. The tones are designed in such a way that unexpected Doppler shifts of the signal to not cause confusion in its correct reception.

Minimum shift keying (MSK) is a form of continuous FSK in which the modulation index is 0.5. Curiously, it can be shown mathematically that MSK can also be written as a four-phase PSK in which the pulse shape is one-half cycle of a sinuoid. For this reason, MSK can be viewed either as having modulation order 2 and symbol duration Tb, or as having modulation order 4 and symbol duration 2Tb.

When an MSK signal is baseband filtered using a Gaussian shape, we obtain Gaus- sian-filtered MSK, or GMSK. A GMSK modulation is parameterized by its BT_b product. A common value of BT_b is 0.3. GMSK is 1.) highly bandwidth efficient; 2.) constant envelope; 3.) less susceptible to nonlinear distortions; 4.) highly power efficient if a trellis receiver is used; 5.) moderately technologically mature; 5.) emerging for use in Deep Space, Space Research and Near Earth missions.

7.1.5 QAM

Quadrature Amplitude Modulation (QAM) signals waveform can be defined as a linear combination of two orthogonal signal waveforms as follows:

tfstfstS )()()( m m 11  m 22

     tftgtf  c  g

CMLP Study Page 45 August 2007 CMLP Final Report

     tftgtf  c  g

 g is the energy of the signal pulse g(t). m = 1,2 ….M.

As is the case for the PSK modulations, QAM can be conveniently viewed as a signal constellation in two dimensions. The points of the constellation on the plane, and the minimum distance between them, determines the basic performance of the modulation. The basic QAM modulations consist of a rectangular grid of points, but modifications of this grid may reduce the average transmission power. The problem with the conventional rectangular QAM modulation is that the corners of the grid are far from the origin. A “rounder” signal constellation can reduce the average power without reducing performance. 32-QAM (cross) and 32-APSK (16/12/4) are examples of this.

QAM is 1.) highly spectrally efficient in a linear power amplifier mode; 2.) non- constant envelope; 3.) highly susceptible to phase noise and nonlinear distortions; 4.) lower power efficiency; 5.) moderately technologically mature. 6.) typically used for high data rate users with a relatively small spectrum allocation. Examples of QAM implementation are 16QAM, 64QAM, and 128QAM.

7.2 Codes

Table 7-2 below lists all candidate coding schemes considered in the CMLP study. Later sections describe the main classes of coding techniques.

CMLP Study Page 46 August 2007 CMLP Final Report

Required Eb/No (dB) with BPSK for BER = Decoding latency (assuming infinitely Gap to Info Encoding fast capacity, for Rate length Code latency decoder) given (r,n), at Code ID Type (r) (k) length (n) (bits) (bits) CWER=1e-4 1E-02 1E-03 1E-04 1E-05 1E-06 1E-07 1E-08 1E-09 1E-10 1 Uncoded Uncoded 1.000 1 1 1 4.32 6.79 8.40 9.59 10.53 11.31 11.97 12.55 13.06 2 CC(3,1/2) Convolutional 0.499 1022 2048 2.22 3.72 4.89 5.85 6.84 7.40 8.04 8.56 9.10 3 CC(5,1/2) Convolutional 0.498 1020 2048 2.03 3.22 4.18 4.99 5.68 6.38 6.95 7.46 7.92 4 CC(7,1/2), delay=5 bits, Q=inf Convolutional 0.500 Inf Inf 10 3.40 5 CC(7,1/2), delay=10 bits, Q=inf Convolutional 0.500 Inf Inf 20 2.77 4.00 6 CC(7,1/2), delay=15 bits, Q=inf Convolutional 0.500 Inf Inf 30 2.39 3.49 7 CC(7,1/2), delay=30 bits, Q=inf Convolutional 0.500 Inf Inf 60 1.88 2.78 3.51 8 CC(7,1/2), delay=60 bits, Q=inf Convolutional 0.499 1784 3574 120 1.70 2.63 3.40 4.18 4.78 9 CC(7,1/2), delay=60 bits, Q=inf Convolutional 0.500 3568 7142 120 1.70 2.63 3.40 4.18 4.78 10 CC(7,1/2), delay=60 bits, Q=inf Convolutional 0.500 8920 17846 120 1.70 2.63 3.40 4.18 4.78 11 CC(7,1/2), delay=60 bits, Q=inf Convolutional 0.500 16384 32774 120 1.70 2.63 3.40 4.18 4.78 12 CC(7,1/2), delay=60 bits, Q=inf Convolutional 0.500 Inf Inf 120 1.70 2.63 3.40 4.18 4.78 13 CC(7,1/2), delay=inf, hard dec. Convolutional 0.500 Inf Inf Inf 3.67 14 CC(7,1/2), delay=inf, Q=3 Convolutional 0.500 Inf Inf Inf 2.13 3.04 3.80 4.46 5.04 5.56 6.02 6.43 6.81 15 CC(7,1/2), delay=inf, Q=8 Convolutional 0.500 Inf Inf Inf 1.91 2.82 3.56 16 CC(7,1/2), delay=inf, Q=inf Convolutional 0.500 Inf Inf Inf 1.70 2.61 3.42 17 CC(7,2/3), delay=60 bits, Q=inf Convolutional 0.667 8920 13380 90 2.41 3.22 3.90 4.57 5.24 18 CC(7,2/3), delay=120 bits, Q=inf Convolutional 0.664 1024 1542 180 2.36 3.19 3.93 4.60 5.23 5.79 6.31 6.79 7.23 19 CC(7,3/4), delay=60 bits, Q=inf Convolutional 0.750 8920 11894 80 2.93 3.76 4.47 5.14 5.77 6.39 20 CC(7,3/4), delay=120 bits, Q=inf Convolutional 0.747 1024 1371 160 2.84 3.66 4.42 5.15 5.78 6.37 6.91 7.40 7.84 21 CC(7,5/6), delay=60 bits, Q=inf Convolutional 0.833 8920 10704 72 3.65 4.45 5.17 5.83 6.44 6.98 22 CC(7,5/6), delay=120 bits, Q=inf Convolutional 0.829 1024 1235 144 3.43 4.24 4.98 5.67 6.32 6.93 7.47 7.96 8.40 23 CC(7,7/8), delay=60 bits, Q=inf Convolutional 0.875 8920 10195 69 4.20 5.06 5.83 6.55 7.25 7.87 24 CC(7,7/8), delay=120 bits, Q=inf Convolutional 0.871 1024 1176 137 3.90 4.67 5.36 6.03 6.68 7.31 7.90 8.44 8.93 25 CC(9,1/2), delay=45?, Q=inf Convolutional 0.500 63 126 90? 1.561 2.28 2.91 3.59 4.12 4.63 5.12 5.57 5.99 26 CC(9,1/2) Convolutional 0.496 1016 2048 2048 1.561 2.28 2.91 3.59 4.12 4.63 5.12 5.57 5.99 27 CC(9,1/2) Convolutional 0.499 4088 8192 8192 1.561 2.28 2.91 3.59 4.12 4.63 5.12 5.57 5.99 28 CC(15,1/4) Convolutional 0.247 1010 4096 4096 0.36 1.01 1.48 2.04 2.55 2.96 3.38 3.80 4.20 29 CC(15,1/6) Convolutional 0.164 1010 6144 6144 0.16 0.77 1.34 1.87 2.42 2.82 3.23 3.64 4.03 30 RS(255,223) Reed-Solomon 0.875 1784 2040 2040 4.78 5.55 5.90 6.16 6.38 6.57 6.74 6.90 7.05 31 RS(255,239) Reed-Solomon 0.937 1912 2040 2040 4.60 5.99 6.46 6.80 7.08 7.33 7.56 7.76 7.96 32 RS(252,220) Reed-Solomon 0.873 1760 2016 2016 4.79 5.56 5.91 6.17 6.39 6.58 6.75 6.91 7.06 33 RS(255,223)+(7,1/2), I=1 Concatenated 0.437 1784 4080 4080 1.97 2.32 2.56 2.78 2.94 G 34 RS(255,223)+(7,1/2), I=2 Concatenated 0.437 3568 8160 8160 2.20 2.34 35 RS(255,223)+(7,1/2), I=3 Concatenated 0.437 5352 12240 12240 1.89 2.14 2.27

36 RS(255,223)+(7,1/2), I=4 Concatenated 0.437 7136 16320 16320 1.89 2.09 2.24 2.35 37 RS(255,223)+(7,1/2), I=5 Concatenated 0.437 8920 20400 20400 1.88 2.08 2.23 2.33 38 RS(255,223)+(7,1/2), I=8 Concatenated 0.437 14272 32640 32640 1.88 2.07 2.19 2.31 2.40 39 RS(255,223)+(7,1/2), I=16 Concatenated 0.437 28544 65280 65280 1.89 2.08 2.20 2.31 2.39 40 RS(255,239)+(7,1/2), I=1 Concatenated 0.469 1912 4080 4080 1.84 2.36 2.72 41 RS(255,239)+(7,1/2), I=2 Concatenated 0.469 3824 8160 8160 1.82 2.21 2.47 2.68 42 RS(255,239)+(7,1/2), I=3 Concatenated 0.469 5736 12240 12240 1.82 2.18 2.37 2.54 43 RS(255,239)+(7,1/2), I=4 Concatenated 0.469 7648 16320 16320 1.83 2.14 2.33 2.49 44 RS(255,239)+(7,1/2), I=5 Concatenated 0.469 9560 20400 20400 1.85 2.13 2.32 2.47 45 RS(255,239)+(7,1/2), I=8 Concatenated 0.469 15296 32640 32640 1.83 2.12 2.30 2.44 2.58 46 RS(255,239)+(7,1/2), I=16 Concatenated 0.469 30592 65280 65280 1.84 2.12 2.30 2.45 2.59 47 RS(255,223)+(7,1/2), I=5 Concatenated 0.469 4780 10200 10200 1.88 2.08 2.23 2.34 48 RS(255,223)+(7,2/3), I=5 Concatenated 0.625 6373 10200 10200 2.62 2.80 2.92 3.03 49 RS(255,223)+(7,3/4), I=5 Concatenated 0.703 7170 10200 10200 3.15 3.34 3.47 3.59 3.68 50 RS(255,223)+(7,5/6), I=5 Concatenated 0.781 7966 10200 10200 3.87 4.07 4.22 4.34 51 RS(255,223)+(7,7/8), I=5 Concatenated 0.820 8365 10200 10200 4.39 4.61 4.77 4.90 52 RS(255,239)+(7,1/2), I=5 Concatenated 0.469 4780 10200 10200 1.85 2.13 2.32 2.47 53 RS(255,239)+(7,2/3), I=5 Concatenated 0.625 6373 10200 10200 2.54 2.82 2.97 3.12 3.25 3.36 54 RS(255,239)+(7,3/4), I=5 Concatenated 0.703 7170 10200 10200 3.06 3.34 3.52 3.66 3.79 55 RS(255,239)+(7,5/6), I=5 Concatenated 0.781 7966 10200 10200 3.79 4.07 4.27 4.42 4.56 56 RS(255,239)+(7,7/8), I=5 Concatenated 0.820 8365 10200 10200 4.33 4.62 4.83 5.00 5.15 Table 7-2: CMLP Coding Schemes

CMLP Study Page 47 August 2007 CMLP Final Report

Required Eb/No (dB) with BPSK for BER = Decoding latency (assuming infinitely Gap to Info Encoding fast capacity, for Rate length Code latency decoder) given (r,n), at Code ID Type (r) (k) length (n) (bits) (bits) CWER=1e-4 1E-02 1E-03 1E-04 1E-05 1E-06 1E-07 1E-08 1E-09 1E-10 57 Turbo(1784,1/6) Turbo 0.166 1784 10728 10728 -0.18 0.00 0.14 0.25 0.34 0.45 0.57 58 Turbo(1784,1/4) Turbo 0.249 1784 7152 7152 0.27 0.42 0.54 0.67 59 Turbo(1784,1/3) Turbo 0.333 1784 5364 5364 0.31 0.51 0.66 0.77 0.88 0.98 60 Turbo(1784,1/2) Turbo 0.499 1784 3576 3576 0.92 1.14 1.29 1.42 1.54 1.67 1.92 61 Turbo(3568,1/6) Turbo 0.166 3568 21432 21432 -0.15 -0.04 0.05 0.14 62 Turbo(3568,1/4) Turbo 0.250 3568 14288 14288 0.13 0.23 0.33 63 Turbo(3568,1/3) Turbo 0.333 3568 10716 10716 0.22 0.36 0.47 0.54 0.62 64 Turbo(3568,1/2) Turbo 0.499 3568 7144 7144 0.83 0.99 1.10 1.19 1.27 65 Turbo(7136,1/6) Turbo 0.167 7136 42840 42840 -0.34 -0.24 -0.17 -0.10 -0.01 66 Turbo(7136,1/4) Turbo 0.250 7136 28560 28560 -0.08 0.03 0.11 0.18 0.26 67 Turbo(7136,1/3) Turbo 0.333 7136 21420 21420 0.16 0.26 0.33 0.39 0.45 68 Turbo(7136,1/2) Turbo 0.500 7136 14280 14280 0.78 0.89 0.97 1.04 1.11 1.16 69 Turbo(8920,1/6) Turbo 0.167 8920 53544 53544 -0.35 -0.27 -0.20 -0.14 -0.10 -0.06 -0.02 70 Turbo(8920,1/4) Turbo 0.250 8920 35696 35696 -0.07 0.03 0.09 0.14 0.19 0.23 0.27 0.42 71 Turbo(8920,1/3) Turbo 0.333 8920 26772 26772 0.14 0.24 0.31 0.37 0.42 0.48 0.58 72 Turbo(8920,1/2) Turbo 0.500 8920 17848 17848 0.77 0.87 0.94 1.01 1.06 1.12 1.30 73 Turbo(16384,1/6) Turbo 0.167 16384 98328 98328 74 Turbo(16384,1/4) Turbo 0.250 16384 65552 65552 75 Turbo(16384,1/3) Turbo 0.333 16384 49164 49164 76 Turbo(16384,1/2) Turbo 0.500 16384 32776 32776 77 BCH-SEC(63,56) BCH 0.889 56 63 63 4.12 5.88 7.07 8.00 8.75 9.39 9.99 10.47 10.95 78 BCH-TED(63,56) BCH 0.889 56 63 63 4.82 7.28 8.90 10.09 11.04 11.80 12.46 13.06 13.55 79 AR4JA(64,1/2) LDPC 0.500 64 128 128 80 AR4JA(1024,1/2) LDPC 0.500 1024 2048 2048 1.14 1.39 1.57 1.74 1.89 2.03 2.17 2.29 2.41 81 AR4JA(1024,2/3) LDPC 0.667 1024 1536 1536 1.89 2.18 2.39 2.59 2.75 2.91 3.04 3.15 82 AR4JA(1024,4/5) LDPC 0.800 1024 1280 1280 2.77 3.08 3.36 3.58 3.76 3.96 4.14 83 AR4JA(4096,1/2) LDPC 0.500 4096 8192 8192 0.93 1.03 1.12 1.19 1.26 1.33 1.39 84 AR4JA(4096,2/3) LDPC 0.667 4096 6144 6144 1.67 1.81 1.90 1.98 2.07 2.13 2.20 85 AR4JA(4096,4/5) LDPC 0.800 4096 5120 5120 2.55 2.72 2.84 2.95 3.04 3.13 3.21 86 AR4JA(16384,1/2) LDPC 0.500 16384 32768 32768 0.75 0.82 0.87 0.91 0.96 1.00 1.03 87 AR4JA(16384,2/3) LDPC 0.667 16384 24576 24576 1.54 1.62 1.68 1.74 1.78 88 AR4JA(16384,4/5) LDPC 0.800 16384 20480 20480 2.47 2.57 2.64 2.70 2.74 89 C2, 50 iterations LDPC 0.875 7136 8160 8160 3.85 4.00 4.19 90 TPC(128,120)^2 Turbo Product 0.879 14400 16384 16384 3.30 3.50 3.59 3.65 3.72 3.78 3.90 91 TPC(64,57)^2 Turbo Product 0.793 3249 4096 4096 2.50 2.80 2.92 3.06 3.17 3.35 3.62 92 TPC(53,46)x(51,44) Turbo Product 0.749 2024 2703 2703 2.30 2.60 2.75 2.90 3.10 3.35 3.65 93 TPC(39,32)^2 Turbo Product 0.673 1024 1521 1521 2.00 2.30 2.55 2.85 3.30 3.60 3.90 94 TPC(32,26)^2 Turbo Product 0.660 676 1024 1024 1.80 2.20 2.50 2.85 3.15 3.55 3.85 95 TPC(19,13)^2 Turbo Product 0.468 169 361 361 1.65 2.25 2.85 3.40 4.10 4.50 5.00 96 TPC(32,26)x(32,26)x(4,3) Turbo Product 0.495 2028 4096 4096 1.50 1.70 1.90 2.10 2.40 2.70 3.10 97 TPC(32,26)x(32,26)x(16,11) Turbo Product 0.454 7436 16384 16384 1.30 1.40 1.46 1.53 1.58 1.65 1.72 98 TPC(16,11)x(16,11)x(16,11) Turbo Product 0.325 1331 4096 4096 0.90 1.08 1.21 1.35 1.50 1.65 1.80 99 TPC(S16xH64^2) Turbo Product 0.744 48735 65536 65536 2.30 2.45 2.50 2.52 2.55 2.57 2.60 100 TPC(H64xH32xS32) Turbo Product 0.701 45942 65536 65536 2.05 2.15 2.20 2.27 2.33 2.40 2.50 101 TPC(H64xH32^2) Turbo Product 0.588 38532 65536 65536 1.80 1.82 1.84 1.86 1.88 2.00 2.04 102 TPC(H16xH64^2) Turbo Product 0.545 35739 65536 65536 1.78 1.80 1.82 1.84 1.86 1.88 2.02 103 TPC(S4xH16xH32^2) Turbo Product 104 TPC(H16^4) Turbo Product 105 TPC(S16xH32^2) Turbo Product 0.619 10140 16384 16384 1.60 1.70 1.80 1.90 2.06 2.22 2.42

106 BCH-LDPC(16200,1/4) BCH-LDPC 0.190 3072 16200 16200 107 BCH-LDPC(16200,1/3) BCH-LDPC 0.323 5232 16200 16200 108 BCH-LDPC(16200,2/5) BCH-LDPC 0.390 6312 16200 16200 109 BCH-LDPC(16200,1/2) BCH-LDPC 0.434 7032 16200 16200 110 BCH-LDPC(16200,3/5) BCH-LDPC 0.590 9552 16200 16200 111 BCH-LDPC(16200,2/3) BCH-LDPC 0.656 10632 16200 16200 112 BCH-LDPC(16200,3/4) BCH-LDPC 0.723 11712 16200 16200

Table 7-2: CMLP Coding Schemes (Continued 1)

CMLP Study Page 48 August 2007 CMLP Final Report

Required Eb/No (dB) with BPSK for BER = Decoding latency (assuming infinitely Gap to Info Encoding fast capacity, for Rate length Code latency decoder) given (r,n), at Code ID Type (r) (k) length (n) (bits) (bits) CWER=1e-4 1E-02 1E-03 1E-04 1E-05 1E-06 1E-07 1E-08 1E-09 1E-10 113 BCH-LDPC(16200,4/5) BCH-LDPC 0.767 12432 16200 16200 114 BCH-LDPC(16200,5/6) BCH-LDPC 0.812 13152 16200 16200 115 BCH-LDPC(16200,8/9) BCH-LDPC 0.879 14232 16200 16200 116 BCH-LDPC(64800,1/4) BCH-LDPC 0.247 16008 64800 64800 117 BCH-LDPC(64800,1/3) BCH-LDPC 0.191 12408 64800 64800 118 BCH-LDPC(64800,2/5) BCH-LDPC 0.397 25728 64800 64800 119 BCH-LDPC(64800,1/2) BCH-LDPC 0.497 32208 64800 64800 120 BCH-LDPC(64800,3/5) BCH-LDPC 0.597 38688 64800 64800 121 BCH-LDPC(64800,2/3) BCH-LDPC 0.664 43040 64800 64800 122 BCH-LDPC(64800,3/4) BCH-LDPC 0.732 47408 64800 64800 123 BCH-LDPC(64800,4/5) BCH-LDPC 0.797 51648 64800 64800 124 BCH-LDPC(64800,5/6) BCH-LDPC 0.831 53840 64800 64800 125 BCH-LDPC(64800,8/9) BCH-LDPC 0.887 57472 64800 64800 126 BCH-LDPC(64800,9/10) BCH-LDPC 0.898 58192 64800 64800 127 Flarion- low threshold LDPC 0.500 4096 8192 8192 0.72 0.86 0.95 1.03 1.11 1.18 >1.43 128 Flarion- low floor LDPC 0.500 4096 8192 8192 0.90 1.03 1.13 1.20 1.29 1.38 1.44 1.51 1.59 129 LDPC 0.750 432 576 576 130 LDPC 0.750 1008 1344 1344 131 LDPC 0.750 1728 2304 2304 132 LDPC 0.500 133 LDPC 0.667 134 LDPC 0.833 135 (3,4,7)LPDC(64) LDPC 0.500 64 128 128 136 (3,4,7)LPDC(128) LDPC 0.500 128 256 256 137 (3,4,7)LPDC(256) LDPC 0.500 256 512 512 138 LDPC 139 F-LDPC (4096, ½) LDPC 0.500 4096 8192 8192 1.25 1.46 1.58 1.70 1.78 1.87 1.95 140 F-LDPC (4096, 2/3) LDPC 0.667 4096 6144 6144 1.90 2.12 2.28 2.38 2.48 2.55 2.62 141 F-LDPC (4096, 4/5) LDPC 0.800 4096 5120 5120 2.75 3.00 3.13 3.25 3.36 3.42 3.52 142 F-LDPC (4096, 8/9) LDPC 0.889 4096 4608 4608 3.50 3.84 4.03 4.18 4.30 4.40 4.68 143 F-LDPC (4096, 16/17) LDPC 0.941 4096 4352 4352 4.10 4.75 4.98 5.15 5.32 5.50 5.90 144 F-LDPC (8192, ½) LDPC 0.500 8192 16384 16384 1.22 1.40 1.50 1.58 1.65 1.70 1.75 145 F-LDPC (8192, 2/3) LDPC 0.667 8192 12288 12288 1.90 2.10 2.18 2.27 2.34 2.41 2.48 146 F-LDPC (8192, 4/5) LDPC 0.800 8192 10240 10240 2.70 2.90 3.05 3.13 3.18 3.25 3.30 147 F-LDPC (8192, 8/9) LDPC 0.889 8192 9216 9216 3.50 3.80 3.90 4.02 4.10 4.17 4.25 148 F-LDPC (8192, 16/17) LDPC 0.941 8192 8704 8704 4.20 4.70 4.90 5.03 5.13 5.28 5.50 149 F-LDPC(16k, ½) LDPC 0.500 16384 32768 32768 1.10 1.19 1.25 1.27 1.32 1.36 1.40 1.44 1.48 150 F-LDPC(16k, 2/3) LDPC 0.667 16384 24576 24576 1.80 1.88 1.94 1.98 2.02 2.06 2.11 2.15 2.18 151 F-LDPC(16k, 4/5) LDPC 0.800 16384 20480 20480 2.65 2.75 2.83 2.88 2.93 2.98 3.03 152 F-LDPC(16k, 8/9) LDPC 0.889 16384 18432 18432 3.50 3.68 3.73 3.78 3.83 3.92 3.97 153 F-LDPC(16k, 16/17) LDPC 0.941 16384 17408 17408 4.20 4.55 4.68 4.75 4.82 4.88 5.00 154 (3,1/2)+acc. SCCC 155 CRC-32 CRC 156 CRC-96 CRC 157 CRC-128 CRC 158 CRC-192 CRC 159 CRC-16-CCITT CRC 160 CRC-16-IBM CRC 161 CRC-16-IEEE CRC 162 SCCC(k=428,1/3) SCCC 0.333 428 1284 1284 163 SCCC(k=428,5/6) SCCC 0.833 428 514 514 164 SCCC(k=428,9/10) SCCC 0.899 428 476 476 165 SCCC(n=16384,1/3) SCCC 0.333 5461 16384 16384 166 SCCC(n=16384,5/6) SCCC 0.833 13653 16384 16384 167 SCCC(n=16384,9/10) SCCC 0.900 14745 16384 16384

Table 7-2: CMLP Coding Schemes (Continued 2)

The CMLP forward error correction coding catalog contains 155 specific codes from nine code groups. Represented code groups are of two broad types: classic (legacy) codes, including convolutional codes (CC), Reed-Solomon (RS) codes, RS+CC concatenated codes, Bose-Chaudhuri-Hocquengham (BCH) codes, and cyclic redundancy check (CRC) codes; and modern iteratively decoded codes, including parallel concatenated convolutional codes (turbo codes), serially concatenated convolutional codes (SCCC), turbo product codes (TPC), and low-density parity-check (LDPC) codes. A general description is provided here for each of these code groups as well as pros, cons and typical applications.

Not explicitly considered by this study is the use of a channel interleaver in conjunction with a code. A channel interleaver breaks up bursts of highly impaired symbols

CMLP Study Page 49 August 2007 CMLP Final Report

by reordering their transmission. Coded symbols are interleaved, transmitted, cor- rupted by potentially long burst errors, and de-interleaved at the receiver. The dein- terleaver causes the decoder to see a more uniform arrangement of impaired symbols. Such techniques have been long established in the Space Network, for example, where a convolutional interleaver, also called a Forney interleaver, is used. The use, or not, of such a channel interleaver is not addressed in this study because it doesn't affect the recommendation of one code over another – all codes of a given length would make use of a channel interleaver in the same way, and all such codes would benefit in the same way. Because all of the coding recommendations of this study are consistent with existing and future designs that either do, or do not, include a channel interleaver, the matter isn't considered further. We do advise, however, that an updated study of the Earth's radio frequency interference (RFI) environment in LEO would prove useful in determining a proper recommendation for whether a channel interleaver is beneficial for those links, and if so, how long the interleaver should be.

7.2.1 Classic Codes

7.2.1.1 Convolutional Codes

Convolutional codes are codes that perform a convolution of the input data stream with the encoder’s impulse responses. Convolutional codes map k bits into n symbols based not only upon the current k information bits but also all previous information bits (or as practical). Convolutional codes can be recursive or non-recursive and systematic or non-systematic.

A convolutional code’s effectiveness is fundamentally limited by the constraint length K of its convolution. Convolutional codes with arbitrarily long constraint lengths can approach the Shannon limit with maximum-likelihood (Viterbi) decoding. Practical convolutional codes are limited to reasonably small constraint lengths (e.g., K = 7), because the decoding complexity increases exponentially with K.

Convolutional codes have been used extensively by NASA and by all of the communications industry. This successful legacy of use and the equipment and infrastructure base it has created is probably the single most important attribute in favor of convolutional codes. Other favorable attributes include: 1) Code synchronization is simple and quick as compared to most block codes, and it can be performed automatically by the decoder without the need to devote additional overhead to a synchronization marker. 2.) These codes have very low latency, on the order of one constraint length for encoding and a handful or two of constraint lengths for decoding. 3) Error rates with Viterbi decoding diminish exponentially with increasing signal- to-noise ratio, and these codes shown no signs of an error floor.

CMLP Study Page 50 August 2007 CMLP Final Report

Attributes of convolutional codes that are not favorable include: 1.) Convolutional codes significantly underperform modern high-performance iteratively decoded codes such as LDPC codes and turbo codes; 2.) Convolutional decoders are not naturally implemented with a parallel architecture, a factor that limits their speed particularly when these codes are used as constituents of iteratively decoded codes (turbo codes and SCCCs).

Convolutional codes are common in all areas of communications including space communications and terrestrial mobile communications. Convolutional codes have been used extensively and very successfully on many NASA missions including the Hubble Space Telescope and Voyager. Two more recent NASA deep space missions, Mars Pathfinder and Cassini, opted for an exceedingly complex-to-decode convolutional code with constraint length K = 15, in order to gain increased performance at relatively low data rates from deep space.

As better performing modern iteratively decoded codes continue to emerge and become commonplace, fewer communication industry areas are choosing convolutional codes over the newer codes. While NASA will continue to support existing convolutional codes for some time, it is expected that there will be some future transition to more efficient codes.

7.2.1.2 Reed-Solomon (RS) Codes

Reed-Solomon (RS) codes are nonbinary systematic codes which introduce (N - K) parity symbols for every K symbols of information. Each RS symbol is formed from multiple bits. RS codes can detect up to (N - K) error symbols, or can correct up to (N - K- e)/2 error symbols in combination with e erased symbols, for any 0 < e < N - K.

Reed-Solomon codes are maximum-distance-separable (MDS), and therefore are capable of correcting the maximum possible number of errors (or combination of errors and erasures) among all codes of given K and N. Furthermore, RS encoders and traditional RS decoders1 are relatively low-complexity and can be implemented in hardware at very high data rates. While this seems to be an ideal combination of code attributes for any application, it is well known that RS codes perform very poorly on many useful channels such as the additive white Gaussian noise (AWGN) channel. There are two primary reasons for this disappointing performance. First, a traditional RS decoder bases its decisions only on hard-limited output from the

1 Recently, enhanced RS decoding algorithms have been developed to utilize soft channel information, but these are high-complexity, low-maturity algorithms that would disqualify RS codes from being considered as “legacy codes”. In fact, these algorithms are of sufficiently low maturity (compared to other newer codes and decoders) that RS codes with enhanced soft decoding were not evaluated for this study in the “modern codes” category either.

CMLP Study Page 51 August 2007 CMLP Final Report

channel, ignoring useful reliability information (worth about 2 dB in AWGN). Sec- ond, there is a channel error magnification effect because each isolated channel symbol error (e.g., a bit error with BPSK modulation) corrupts a larger nonbinary RS symbol. Together these two effects account for several dB of performance loss when RS codes are applied to soft-output channels such as AWGN.

RS codes can be efficient when applied to channels that are inherently bursty and produce hard-limited output. An RS code can also be useful for a soft-output channel afflicted with white noise if it is concatenated (as an outer code) with an inner code (such as a convolutional code) that is more suitable for exploiting the characteristics of the channel corruptions (see next subsection).

RS codes with a symbol size of 8 bits and (N, K) = (255, 223) and (255, 239), as well as shortened versions of these codes, are currently supported by GN, SN and DSN.

7.2.1.3 RS+CC Concatenated Codes

A concatenation of a Reed-Solomon outer code with a convolutional inner code (RS+CC concatenated code) is a classic code that exploits the MDS properties and the large block size advantages of the RS outer code, together with the ability of the convolutional inner decoder to efficiently extract soft information from the channel with low complexity. The performance of RS+CC concatenated codes is characterized by a steeply falling error rate curve. The slope of this curve is ideally determined by the block size of the RS outer code, which is nearly two thousand information bits for classic RS codes using 8-bit symbols, and for useful shortened versions of these codes. The burstiness of errors output from the inner Viterbi decoder greatly reduces the error magnification effect that ordinarily would diminish the effectiveness of an RS code with 8-bit symbols if it were connected directly to a white noise channel. However, because the Viterbi decoder error bursts are unpredictable in length, and occasionally longer than several RS symbols, classic RS+CC concatenations generally include a block interleaver between the inner and outer code in order to break up long Viterbi decoder bursts into smaller pieces distributed among multiple RS codewords. This interleaving improves the RS+CC concatenated code’s performance, but at the expense of increasing its block size and therefore its latency. The overall information block size is nearly nine thousand bits for the classic RS+CC concatenated code built from an 8-bit RS outer code, a K = 7 convolutional inner code, and a depth-5 interleaver.

The deteriorated performance of an RS+CC convolutional code without interleaving, as compared to the same code with sufficient interleaving to break up the Viterbi decoder error bursts, is similar in principle to the error floor phenomenon that pla- gues turbo codes at very low error rates. The longer the constraint length (and hence longer error bursts) of the convolutional code, and the smaller the size of the RS

CMLP Study Page 52 August 2007 CMLP Final Report

code, the more the performance of the RS+CC concatenated code will be determined by that of its inner convolutional code and not by the complementary strengths of the overall concatenation. However, this degradation of performance shows up as a fairly uniform diminishment of slope of the concatenated code’s error rate curve, unlike the sharp transition in slope in the performance curve of a turbo code at the beginning of its error floor region. Furthermore, because the typical error bursts for a K = 7 convolutional code are still small compared to the size of an 8-bit RS code, the performance curve of this particular RS+CC concatenation without interleaving is still much steeper than that of a turbo code past the start of its error floor.

Even with ideal (infinitely long) interleaving, RS+CC concatenated codes are not capable of approaching the Shannon capacity limit of performance more closely than about 2 dB, unless the inner convolutional code’s constraint length is impractically long or the combined decoding of the inner and outer code is impractically complex. A highly complex RS+CC concatenated code was designed to support NASA’s Gali- leo mission to Jupiter after Galileo was forced to transmit exclusively through its low-gain antenna at extremely low data rates (around 100 bps). This code featured a very long constraint-length convolutional inner code (K = 14), a variable-rate RS outer code, a long interleaver (depth-8), and four stages of alternating decoding between the inner and outer codes (a form of iterative decoding passing hard rather than soft information during the iterations). With these enhancements (generally impractical except at Galileo’s low data rates), the gap to capacity was shaved to about 1 dB. But this performance still falls about 1/2 dB short of that of modern turbo or LDPC codes of similar sizes and rates and much lower complexity.

As with all long block codes, RS+CC concatenated codes require accurate synchronization to determine the starting position of a codeword from among thousands of possible locations. As compared to similar synchronization requirements for other long block codes (including turbo codes, TPCs, SCCCs, and LDPC codes), synchronization for RS+CC concatenated codes can be somewhat less burdensome due to the inner convolutional decoder’s self-synchronizing capability and the decoupling of the inner and outer decoders. This allows codeword synchronization for RS+CC concatenated codes to be accomplished using Viterbi decoded bits rather than a larger number of lower-reliability raw channel symbols. In this case, synchronization performance is not necessarily improved, but the required processing is simplified.

RS+CC concatenated codes are currently widely used for GN, SN and DSN.

7.2.1.4 BCH Codes

Bose-Chaudhuri-Hocquengham (BCH) codes are classic binary codes that can be designed to correct small to moderate numbers of bit errors without excessive encoding or decoding complexity. As with many other classic codes, reasonable-

CMLP Study Page 53 August 2007 CMLP Final Report complexity decoding algorithms for BCH codes use hard inputs. For this reason, BCH codes are generally a poor choice to apply directly to a soft-output channel such as a Gaussian noise channel. Additionally, BCH codes have generally taken a back seat to RS codes as an outer code in a concatenated system, due to an RS code’s capability to correct more bit errors when both types of codes are constrained to similar complexities.

In modern times, BCH codes are used in the DVB-S2 standard as an outer code to an inner LDPC code. The purpose of the BCH code in this setting is to lower an otherwise unacceptably high error floor due to frequent errors of very low weight produced by the LDPC decoder. For the purposes of the CMLP study, BCH codes were evaluated only in the context of this specialized application, in conjunction with the particular LDPC codes designed for the DVB-S2 standard. More recent designs of LDPC codes, such as C2 and the AR4JA family, are not susceptible to the low-weight error events that would be correctable by a BCH code. In this case, concatenation of the LDPC code with a BCH code would be superfluous and would only serve to reduce the power efficiency of the overall code without any appreciable lowering of its error rate.

7.2.1.5 CRC Codes

Cyclic redundancy check (CRC) codes are codes that append a fixed number of parity bits to large information blocks of varying lengths for the purposes of error detection only. Typical CRC codes use 16-bit, 32-bit, or somewhat longer parity sequences. A CRC code is usually used as an outer code concatenated with an inner error-correcting code (which may itself be a concatenation of two or more constituent codes).

To first order, the probability that an erroneous codeword escapes detection by an outer CRC code is roughly the same as the probability that its parity bits agree with an equal number of random bits. The conditional undetected error probability of an m-bit CRC code is roughly 2–m if the inner codeword’s typical error patterns are long and varied. Good CRC codes are generally designed to offer guaranteed detection of small numbers (e.g., up to 3) of bit errors, on the assumption that higher-weight error patterns occur with much lower probability. This design feature greatly en- hances a CRC code’s detection performance with uncoded data, but it is nearly worthless when used with a powerful inner error-correcting code that is likely to make either no bit errors or many bit errors in bunches.

Because CRC codes are not used to correct any errors, they cannot improve the ability of the overall coding system to approach the Shannon limit of performance. In fact, the power efficiency of the overall code is reduced by the rate of the CRC code, with no improvement in error-correcting capability. When the CRC code is attached

CMLP Study Page 54 August 2007 CMLP Final Report

to large inner codewords or data frames thousands of bits long, the CRC code’s overhead penalty is very small and is tolerated in return for its ability to reliably detect errors. However, the performance penalty for using a CRC code becomes non- negligible if it is used to protect smaller codewords or frames on the order of a few hundred bits or less.

Encoding and decoding of CRC codes is accomplished using simple linear feedback shift registers, and this encoding and decoding architecture does not change with the size of the frame being protected. This invariance of coding and decoding architecture to code block size is a primary reason why CRC codes are generally preferred over other codes with equal or better error detection capabilities. Often, however, the inner error-correcting code has error detection capabilities of its own, and in such cases the CRC code becomes a useless appendage that reduces power efficiency without offering improved error detection. A CRC code is not needed to improve the inherent error detection capabilities of the classic (255, 223) RS code (or the corresponding RS+CC concatenated code), and CRC codes up to at least 32-bits do not appear to improve the native error detection capabilities of well-designed LDPC codes, such as C2 or the AR4JA family.

7.2.2 Modern Codes

7.2.2.1 Turbo Product Codes (TPC)

A product code is obtained from constituent (N1, K1) and (N2, K2) codes by filling an N1 by N2 rectangular array of coded bits with: 1) a K1 by K2 rectangular array of information bits; 2) a K1 by (N2 – K2) rectangular array of parity bits computed by applying the (N2, K2) code’s encoding rule to each of the K1 rows of information bits; and 3) an (N1 – K1) by N2 rectangular array of parity bits computed by applying the (N1, K1) code’s encoding rule to each of the K2 columns of information bits and (N2 – K2) columns of parity bits computed in the previous step. This product code maps K1K2 information bits into a total of N1N2 coded bits. An identical product code results if the column encoding is done first and the row encoding second. If one of the constituent codes is itself a product code, the resulting code is a product code in three or more dimensions.

Product codes are classic codes that have not found many useful applications until recently, due to poor minimum distance for relatively large block size and the un- availability of soft maximum-likelihood decoding algorithms. However, the iterative decoding revolution launched by turbo codes also sparked a revival of product codes. Product codes can be decoded iteratively by alternating the decoding of rows and columns, and passing soft extrinsic information between the row and column decodings in the manner of turbo decoding. A classic product code decoded in this

CMLP Study Page 55 August 2007 CMLP Final Report manner is called a turbo product code (TPC), and is now regarded as a modern code.

Encoders and decoders for turbo product codes can be implemented at high speed, because the individual row and column encodings and decodings can be performed in parallel. Typical constituent codes of a TPC are single-parity-check codes, Ham- ming codes, and extended Hamming codes, with small minimum distances of 2, 3 and 4, respectively. Such constituents are selected because they are fairly easy to decode individually, and their relatively high rates keep the rate of the product code (equal to the product of its constituents’ rates) from being unreasonably low. How- ever, since the minimum distance of the product code is the product of its constituents’ minimum distances, two-dimensional product codes built from such constituents will include many incorrect codewords within distance 16 of the true codeword, and the decoder’s error rate will reach a low-slope error floor region where further improvements are limited by the difficulty of distinguishing among these relatively close neighbors.

TPCs in three dimensions built from extended Hamming constituent codes can achieve a minimum distance of 64, and their error rate curves fall off much more steeply even when the minimum-distance neighbors dominate the performance. As a result, their error floors may be imperceptible in many applications. However, TPCs of three (or more) dimensions are more complex, their iterations require an extra round (or more) of decoding, their overall rates are fairly low, and their overall block sizes are quite large.

TPCs share many of the same features as other modern codes including turbo codes, SCCCs, and LDPC codes. As long block codes, they have long latency relative to classic convolutional codes, and they require accurate synchronization among thousands of possibilities to determine the starting location of a codeword. As long codes that can achieve near-maximum-likelihood performance via iterative decoding, some TPCs can approach the Shannon performance limit as closely as turbo and LDPC codes, particularly for very large block sizes on the order of tens of thousands of bits. Generally speaking, turbo, SCCC and LDPC codes offer much more flexibility for designing near-optimal codes at a wide range of rates and sizes down to a thousand bits or lower, while only a few scattered point designs of TPCs are equally near-optimal unless the size of the code is extremely large.

7.2.2.2 Turbo Codes

Turbo codes are parallel concatenations of two or more simple recursive convolutional codes, used to encode differently permuted versions of the same information sequence. The different permutations of the input information bits are accomplished by one or more interleavers. Turbo codes are decoded iteratively by passing soft ex-

CMLP Study Page 56 August 2007 CMLP Final Report

trinsic information between two relatively simple convolutional decoders tasked to decode the constituent codes separately. If the information block is reasonably large and the interleaver(s) sufficiently random, the iterative turbo decoder achieves nearly the same performance as an impossibly complex maximum-likelihood decoder for the same code. Furthermore, turbo codes can approach the Shannon capacity limit of performance with well-designed constituent codes and interleaver(s).

Unlike turbo product codes, which are a classic code structure to which iterative decoding principles are applied, turbo codes (as well as serially concatenated convolutional codes and LDPC codes) are modern codes that were designed from the start to be decoded iteratively. For example, the recursive property of the turbo code’s constituent convolutional codes is critically important for its near-optimal performance, but this property has only a minor impact on the performance of these same convolutional codes decoded classically.

Practical turbo codes are generally limited to two constituent codes, because of the need for multiple interleavers, the increased length of an iteration cycle, and the lower rate of the overall turbo code, when more than two constituents are used. However, parallel concatenations of two constituent codes are susceptible to an error floor, where the near-optimal performance of the overall turbo code breaks down and further reductions in error rate are limited by the properties of the weak constituent codes. With good choices of constituent codes and interleaver, this error floor can be driven low enough for many applications, e.g., CWER in the range of 10–6 to 10–8 and BER an order of magnitude lower, but not sufficiently low for applications that require error rates a few orders of magnitude lower than this.

Even when limited to two constituents, the natural rates of parallel concatenations without puncturing are 1/3 and lower. Higher turbo code rates can be produced by puncturing some of the constituent decoders’ outputs, but excessive puncturing can be detrimental to performance. Most useful turbo codes have been developed for code rates 1/2 and lower.

Good constituent codes for turbo codes have very short constraint lengths (e.g., K = 3 to 5), which makes them even simpler to decode than a classic medium-constraint- length convolutional code with K = 7 (even allowing for the turbo decoder’s requirement that its constituent decoders produce soft rather than hard outputs). However, the turbo decoder’s overall complexity is much higher than that of the K = 7 convolutional code, due to its needs for two such constituent decoders, for performing multiple iterations, for processing a large block of data at once, and for interleaving and deinterleaving the soft outputs from each constituent decoder during the course of each iteration. On the other hand, the complexity of turbo decoding is significantly lower than that of Viterbi decoding of the long-constraint-length (K = 15) classic convolutional code used by Mars Pathfinder and Cassini.

CMLP Study Page 57 August 2007 CMLP Final Report

Turbo codes need to encode reasonably large blocks of information (e.g., a thousand or more bits) in order to achieve near-optimal performance commensurate with their block sizes. This contributes to high latency, and a need for accurate code block synchronization as discussed previously for TPCs.

Turbo codes are currently in use on NASA’s Mars Reconnaissance Orbiter (MRO) and are supported by the DSN.

7.2.2.3 Serially Concatenated Convolutional Codes (SCCC)

A serially concatenated convolutional code (SCCC) is a serial concatenation of two codes similar in concept to the classic RS+CC concatenation. The inner and outer codes of an SCCC are both short-constraint-length convolutional codes, typically K = 3 for the outer code and K = 3 to 5 for the inner code. The SCCC’s inner convolutional code is recursive (as are the parallel constituents of turbo codes). Between the inner and outer codes is an interleaver that resembles the random-like interleaver of turbo codes rather than the regular rectangular interleaver of RS+CC concatenated codes.

As with turbo codes, an SCCC is a modern code structure that was never considered useful or practical until iterative decoding algorithms were developed to decode it effectively and with reasonable complexity. A turbo code can be regarded as a special type of SCCC with a simple repetition code replacing the SCCC’s outer convolutional code and its interleaver obeying some additional constraints.

The outer code of an SCCC generally has minimum distance at least 3, and this property eliminates the appearance of error floors at error rate levels typical of the error floors of turbo codes constructed from only two constituents. Furthermore, the outer convolutional code achieves this higher minimum distance without lowering the overall code rate as much as a turbo code’s rate is lowered when it has more than two parallel constituents. Also, the serial combination of two non-trivial codes offers greater flexibility for puncturing either or both constituents to achieve higher rates while not sabotaging the near-optimality of the code’s performance.

On the other side of the ledger, it is difficult to design SCCCs with decoding thresholds as low as those of turbo codes, and their decoding complexity is somewhat higher.

7.2.2.4 Low-Density Parity-Check (LDPC) Codes

LDPC codes are old codes but not classic. Having been invented by Gallager nearly a half-century ago, they lay dormant for many decades until similarities were noted between Gallager’s code constructions and iterative decoding methods, and those of

CMLP Study Page 58 August 2007 CMLP Final Report

Berrou et al in their more recent invention of turbo codes. Modern LDPC codes have been re-engineered and optimized in many directions over the past decade since their rediscovery.

An LDPC code is defined by a sparse parity-check matrix containing only a few 1s in each row and column. This parity-check matrix can be represented by a sparsely connected graph introduced by Tanner. The Tanner graph also describes the paths along which messages are passed when the LDPC code is decoded iteratively. There are individual nodes in the Tanner graph for all of the coded bits and code constraints, and this feature enables extreme parallelization of the decoder’s operations within each iteration. This contrasts with the time-sequential forward and backward message passing that takes place within each iteration on the trellis graph that represents the constituents of a turbo code or SCCC. This massive inherent paral- lelizability is a major advantage of the LDPC decoding algorithm, allowing LDPC decoding speeds to be limited mainly by the amount of hardware that can practi- cally assembled to perform primitive message passing operations in parallel.

In addition to providing the best potential for achieving high decoding speeds among iteratively decoded codes, LDPC codes also offer more degrees of design freedom compared to turbo codes and SCCCs and especially TPCs. This has enabled LDPC code designers to trade off decoding threshold, error floor performance and other attributes more effectively than for these other modern codes. Specific LDPC codes have been designed to approach microscopically close to the Shannon limit of performance, and there is no theoretical limitation on how low their error floors can be pushed. A decade of improving LDPC code design methods has resulted in codes of a wide range of practical sizes and rates that perform reasonably close to the Shannon limit down to error floor levels that are virtually undetectable.

Early LDPC code designs were highly unstructured, because random-like connections in the Tanner graph provide a statistically sound method to generate ensem- bles of good LDPC codes. However, unstructured designs lead to impractical decoding, due to the difficulty of properly routing messages in a large randomly connected (albeit sparsely connected) graph, notwithstanding the fact that the number of computations needed to generate each message does not increase if the connections are unstructured. More recently, quasicyclic LDPC codes have been designed from a small template graph (protograph) and a selection of circulant permutations. The Tanner graphs of quasicyclic LDPC codes have more regularly structured connections that simplify the LDPC decoder’s architecture.

Despite their many intrinsic advantages, LDPC codes have not totally displaced turbo codes. The realm of rates less than 1/2 where turbo codes work best is also where good LDPC code designs become more difficult. At low rates, an LDPC code’s Tan- ner graph acquires more connections as additional parity-check constraints are add-

CMLP Study Page 59 August 2007 CMLP Final Report

ed. Furthermore, iterations take a very long time to converge, since each input symbol from the channel has very low reliability due to its highly diluted signal-to-noise ratio. In contrast, iterations proceed more quickly for low-rate turbo codes or SCCCs, since their constituent convolutional decoders use aggregated messages from several weak channel symbols to label the branches of their decoding trellises before starting their iterations. Finally, turbo codes and SCCCs have an advantage over LDPC codes in their inherent ease of encoding. An LDPC code is defined via its sparse parity-check matrix, but the corresponding generator matrix for encoding the code is not sparse. This is in contrast to turbo codes and SCCCs, with their constituent short-constraint-length (low-density) convolutional encoders. However, recent structured LDPC code designs, especially quasicyclic LDPC codes, have made high-speed encoders feasible for LDPC codes as well.

7.3 Multiple Access Schemes

The candidate multiple access schemes are described briefly in this section.

7.3.1 General Schemes 7.3.1.1 Demand Assigned Multiple Access (DAMA) Demand Assigned Multiple Access (DAMA) is a technique that assigns network resources in an as-needed manner based upon user requests. This MA technique enables a large pool of users to efficiently use the network by taking advantage of the fact that not all users will require access to the network at the same time. When a user no longer requires a data link to the network, that link resource is freed up for use by another user.

The primary benefit of DAMA is that network efficiency can be dramatically improved over conventional MA techniques (e.g., FDMA and TDMA) for most user traffic types, especially low duty cycle traffic. The drawbacks of DAMA are 1.) efficiency collapses when all users are trying to access the network at once, for example, in an emergency situation; 2.) data latency and latency variability are increased; 3.) efficiency is a function of distance between central network station and users.

Due to DAMA’s superior performance for low duty cycle users, this MA technique is often used in VSAT communication systems.

7.3.1.2 Space Division Multiple Access (SDMA) Space Division Multiple Access (SDMA) allows the reuse of both time and frequency through parallel channels created through directional antennas. SDMA algorithms

CMLP Study Page 60 August 2007 CMLP Final Report

operate by transmitting different signals simultaneously on different transmit antennas at the same frequency and by using multiple receive antennas for decoding. SDMA can be combined with a multiple access scheme like TDMA, CDMA, or OFDM. One SDMA algorithm is Maximum Likelihood Decoding (MLD) which is a method that compares the received signal with all possible transmitted vectors and estimates the actual signal according to the Maximum Likelihood principle.

SDMA exploits multipath scattering to improve system capacity and BER performance. The primary disadvantage of SDMA, especially MLD, is that its complexity grows exponentially with the number of transmit antennas.

SDMA is a relatively new concept that has been demonstrated in indoor laboratory environments at realistic SNR’s and BER’s. In addition, several companies aim at SDMA commercial products that support certain multiple access schemes.

7.3.2 Time Sharing Schemes 7.3.2.1 Time Shared The time shared Multiple Access (MA) technique is a schedule-based time sharing approach, where one user accesses the network at a time for a unique, user- requested duration. Time shared multiple access is a very mature and simple MA technique.

Many limited availability assets utilize a time shared MA approach. A typical application of the time shared MA technique is the TDRSS Single Access communication services.

7.3.2.2 Time Division Multiplexing (TDM) Time division multiplexing (TDM) is a technique where data from a fixed set of active users is multiplexed into one data stream which is transmitted over the channel. Individual data streams are regenerated at the receiving end using a demultiplexer.

Users can provide data at their unique data rate and the data multiplexer will buffer data as necessary. TDM is a very mature and simple MA technique. A typical application of TDM is a T1 line within the Public Switched Telephone Network (PSTN).

7.3.2.3 Time Division Multiple Access (TDMA) Time division multiple access (TDMA) is a technique where a fixed set of users access a single radio-frequency (RF) channel without interference by allocating unique time slots to each user and allowing only one user to either transmit or receive during that time slot. Each user utilizes a cyclically recurring time slot. The transmission

CMLP Study Page 61 August 2007 CMLP Final Report

is in a buffer-and-burst manner, thus the transmission for any user is non- continuous.

TDMA is a mature MA technique of moderate complexity. The drawbacks of TDMA are 1.) requires time synchronization across network; 2.) requires the user hardware to support the burst data rate rather than the individual user data rate; 3.) has a hard limit on the number of users that can be supported; 4.) becomes inefficient when user data is not evenly recurring; 5.) suffers disproportionately from multipath / fading as compared to certain other MA techniques; 6.) spectral efficiency limited by need for time guard bands.

TDMA is a popular MA techniques used in a wide array of applications including 2G cellular phone systems.

7.3.3 Frequency Sharing Schemes 7.3.3.1 Frequency Division Multiple Access (FDMA) – General Frequency Division Multiple Access (FDMA) is a technique where the allocated RF bandwidth is divided into channels and these frequency channels are uniquely assigned to a fixed set of users. The channelization plan ensures sufficient bandwidth is available to support the unique data rate of the assigned user and to ensure adjacent channel interference is at acceptable levels.

FDMA is a very mature MA technique and is of low complexity. The drawbacks of FDMA are 1.) the aggregate FDMA signal can have a high peak to average power ratio – which translates into a potential need for a backed-off Power Amplifier (PA) with a large linear dynamic range or multiple PA’s; 2.) has a hard limit on the number of users that can be supported; 3.) becomes inefficient when user data is not evenly recurring; 4.) user RF hardware is unique from user to user due to different operational center frequencies; 5.) suffers disproportionately from multipath / fading as compared to certain other MA techniques; 6.) spectral efficiency limited by need for frequency guard bands.

FDMA is a popular access method used in many communication systems including satellite communication systems.

7.3.3.2 Orthogonal Frequency Division Multiplexing (OFDM) Orthogonal Frequency Division Multiplexing (OFDM) is a form of FDMA which utilizes a large number of closely-spaced orthogonal sub-carriers. Each sub-carrier is modulated with a conventional modulation scheme (such as QPSK or QAM) at a low symbol rate.

CMLP Study Page 62 August 2007 CMLP Final Report

OFDM can be a very spectrally efficient MA technique which is resilient to multipath / fading. A primary drawback of OFDM is that the transmitter and receiver must meet stringent oscillator (carrier and data clock) fidelity requirements.

OFDM is a rapidly-emerging, moderately mature MA technique used in DVB-T, WiFi (IEEE 802.11) and WiMax (IEEE 802.16) systems.

7.3.3.3 Wavelength Division Multiple Access (WDMA) Wavelength Division Multiple Access (WDMA) is an optical version of FDMA. In a WDMA system, prisms are used to multiplex and de-multiplex the optical signals. WDMA shares the same characteristics as FDMA. WDMA is commonly used in fiber optical network applications.

7.3.4 Direct Sequence Techniques 7.3.4.1 Traditional Direct Sequence Spread Spectrum Code Division Multiple Access (DSSS CDMA) Traditional Direct Sequence Spread Spectrum (DSSS) Code Division Multiple Access (CDMA) is a technique that utilizes orthogonal pseudo-random noise (PN) spreading codes uniquely assigned to users to enable use of the entire allocated bandwidth simultaneously by all users. The receiver performs a time correlation operation to detect only the specific desired PN code. All other user signals appear as noise due to decorrelation.

A benefit of CDMA is that users are only required to transmit at their unique data rate, not the aggregate burst rate. CDMA is resilient to multipath / fading, and there is not a hard limit on the maximum number of users which can be supported.

However, CDMA involves an increased transmitter and receiver complexity relative to TDMA and FDMA, the presence of multiple access noise, and the near-far problem – one extreme of the MA noise issue when one signal dominates all others.

CDMA is a very mature and popular access method used in a wide array of applications including mobile phone systems and satellite communication systems.

7.3.4.2 Constant Envelope DSSS CDMA When DSSS CDMA waveforms of comparable power and spectrum occupancy are superimposed, the summed signal tends to take on an envelope that is normally distributed with a zero mean and a variance reflecting the average power of the signals. If these signals are transmitted from a common source, a highly linear (Class A) high power amplifier (HPA) is required to transmit the sum signal without distortion.

CMLP Study Page 63 August 2007 CMLP Final Report

Constant envelope CDMA is a multiplexing algorithm that eliminates the need for linear amplification by achieving 0 dB of peak-to-average power ratio (PAPR). A number of equal-rate, chip-synchronous codes are nonlinearly multiplexed into a single representative code that is transmitted as a constant envelope signal. The input power distribution of the codes may be arbitrary. The benefit is that a saturating (Class C) HPA may be used due to the constant power envelope.

The primary drawback is that situation-dependent latency occurs because users are required to relay signals to a common node for multiplexing and transmission, which also rules out the use of this technique for node-to-base (return) links. This technique also incurs a typical multiplexing loss of 1.0 to 2.0 dB

Constant envelope CDMA is applied to GPS Modernization satellite payloads, possibly the first application.

7.3.4.3 Frequency Hopped Spread Spectrum (FHSS) Frequency Hopped Spread Spectrum (FHSS) is one of the two types of non-hybrid spread spectrum techniques, the other being direct-sequence spread spectrum (DSSS). FHSS is a transmission technology where the data signal is modulated with a narrowband carrier signal that "hops" in a random but predictable sequence from frequency to frequency as a function of time over a wide band of frequencies. The signal energy is spread in frequency domain rather than chopping each bit into small pieces in the time domain. The pseudorandom change of the carrier frequencies of the user randomizes the occupancy of a specific channel at any given time, therefore allowing for multiple access over a wide range of frequencies.

The transmission frequencies are determined by a spreading, or hopping code. The receiver must be set to the same hopping code and must listen to the incoming signal at the right time and correct frequency in order to properly receive the signal. Current FCC regulations require manufacturers to use 75 or more frequencies per transmission channel with a maximum dwell time (the time spent at a particular frequency during any single hop) of 400 ms.

FHSS has no hard limit on number of users, it is resilient to multipath/fading, and narrow band interference is mitigated without the need for selective filtering as long as some form of coding redundancy is present.

FHSS has similar drawbacks as DSSS CDMA.

FHSS is a mature and popular access method used in a wide array of applications including mobile phone systems and sat communication systems, and in military systems.

CMLP Study Page 64 August 2007 CMLP Final Report

7.3.4.4 Direct Sequence/Frequency Hopped Spread Spectrum (DS/FHSS) Direct Sequence/Frequency Hopped Spread Spectrum (DS/FHSS) combines DSSS and FHSS. The carrier frequencies of direct sequence modulated signals are made to hop periodically in a pseudorandom fashion.

DS/FHSS has benefits similar to traditional DSSS and FHSS, however it also miti- gates the near-far problem. The drawbacks are that more bandwidth is required than for DSSS and FHSS, and receiver complexity is increased.

It is used for cellular phone to avoid near-far effect in a cell.

7.3.5 Random Access Methods 7.3.5.1 Random Access - Pure ALOHA The pure ALOHA is a protocol for satellite and terrestrial radio transmissions. In pure Aloha, a user can access a channel as soon as a message is ready to be transmitted. After a transmission, the user waits for an acknowledgement on either the same channel or a separate feedback channel. In case of collision, the terminal waits for a random period of time and retransmits the message. As the number of users increases, a great latency occurs because the probability of collision increases.

Pure ALOHA is a mature technique. It is easy to implement, and there is no hard limit on number of users.

However, pure ALOHA has poor spectral efficiency and latency due to collision and retransmission, it relies on data link layer protocol to achieve high performance, and its performance degrades dramatically with an increase in users.

ALOHA was the basis for Ethernet, a Local Area Network (LAN) protocol.

7.3.5.2 Random Access - Slotted ALOHA In Slotted ALOHA, time is divided into equal time slots of length greater than the packet duration. The subscribers have synchronized clocks and transmit a message only at the beginning of a new time slot, thus results in a discrete distribution of packets and prevents partial collisions where one packet collides with a portion of another.

Slotted ALOHA has very similar characteristics as pure ALOHA. The primary benefit is better throughput vs. number of users because of the collision reduction. The drawback is slightly higher complexity.

CMLP Study Page 65 August 2007 CMLP Final Report

Slotted ALOHA is popular for LANs.

7.3.5.3 Random Access - Reservation ALOHA Reservation ALOHA is a packet access technique, evolved from pure ALOHA, implemented based upon time division multiplexing. In this protocol, certain packet slots are assigned with priority, and it is possible for users to reserve slots for the transmission of packets. Slots can be permanently reserved or reserved on request. For high traffic conditions, reservations on request offers better throughput.

Typically there are fewer time slots than users by assuming low probability of simultaneous transmission of all users. Reservation ALOHA reduces collision probability and latency, and there is no hard limit on the number of users.

Reservation ALOHA involves higher complexity than pure ALOHA. It relies on more complicated data link layer protocol support; and its performance degrades dramatically with the increase in the number of users. Reservation ALOHA is popular for LANs.

7.3.5.4 Random Access - Packet Reservation Multiple Access (PRMA) PRMA was originally proposed to integrate bursty data and human speech. It uses a discrete packet time technique similar to reservation ALOHA and combines the cy- clical frame structure of TDMA in a manner that allows each TDMA time slot to carry either voice or data, where voice is given priority. Within each frame, there are a fixed number of time slots that may be designated as either “reserved” or “available”, depending on the traffic as determined by the controlling station.

Compared to pure ALOHA, the PRMA reduces collision probability and latency but its primary drawback is slightly higher complexity.

Its application is primarily for voice and bursty data.

7.3.5.5 Random Access - Carrier Sense Multiple Access (CSMA) Pure ALOHA does not listen to the channel before transmission, and therefore does not exploit information about the other users. CSMA is similar to pure ALOHA but having the capability of listening to the channel. CSMA protocols are based on the fact that each terminal on the network is able to monitor the status of the channel before transmitting information. If the channel is idle (i.e., no carrier is detected), then the user is allowed to transmit a packet based on a particular algorithm that is common to all transmitters on the network.

CMLP Study Page 66 August 2007 CMLP Final Report

CSMA includes no hard limit on the number of users, it increases the transmission efficiency, and it reduces collision probability and latency.

The primary drawback is that CSMA relies on more complex data link layer protocol support than pure ALOHA.

It is a mature technique and popular for LANs.

7.3.5.6 Random Access - Carrier Sense Multiple Access with Collision Detection (CSMA/CD) CSMA/CD is CSMA with collision detection (CD). A user monitors its transmission for collisions. If two or more terminals start a transmission at the same time, collision is detected, and the transmission is immediately aborted in midstream. This is handled by user having both a transmitter and receiver that is able to support listen- while-talk operation.

The primary benefit of CSMA/CD is that it further reduces collision probability and improves the spectral efficiency and latency performance for large number of users as compared to CSMA.

The primary drawback is that it has slightly higher complexity than CSMA. It is a mature technology and very popular for LANs.

7.3.5.7 Random Access - Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) CSMA/CA is a modification of pure Carrier Sense Multiple Access (CSMA). It is similar in many ways to CSMA/CD. Collision avoidance is used to improve the performance of CSMA by attempting to be "less greedy" on the channel. If the channel is sensed busy before transmission then the transmission is deferred for a "random" interval. This reduces the probability of collisions on the channel.

CSMA/CA is used where CSMA/CD cannot be implemented due to the nature of the channel. One of the problems of wireless LANs is that it is not possible to listen while sending, therefore collision detection is not possible. Another reason is the hidden terminal problem, whereby a node A, in range of the receiver R, is not in range of the sender S, and therefore cannot know that S is transmitting to R.

CSMA/CA can optionally be supplemented by the exchange of a Request to Send (RTS) packet sent by the sender S, and a Clear to Send (CTS) packet sent by the intended receiver R, alerting all nodes within range of either the sender, the receiver,

CMLP Study Page 67 August 2007 CMLP Final Report

or both, to keep quiet for the duration of the main packet. This is known as the IEEE 802.11 RTS/CTS exchange.

CSMA/CA is currently used by WiFi networks. 7.3.5.8 Random Access - Multiple Access with Collision Avoidance (MACA) MACA is a slotted media access control protocol used in wireless LAN data transmission to avoid collisions caused by hidden station problem and to simplify ex- posed station problem.

The basic idea of MACA is a wireless network node makes an announcement before it sends the data frame to inform other nodes to keep silent. When a node wants to transmit, it sends a signal called Request-To-Send (RTS) with the length of the data frame to send. If the receiver allows the transmission, it replies the sender a signal called Clear-To-Send (CTS) with the length of the frame that is about to receive. Meanwhile, a node that hears RTS should remain silent to avoid conflict with CTS; a node that hears CTS should keep silent until the data transmission is complete.

The primary benefit of MACA is that it reduces the probability of collisions. How- ever, MACA relies on data link layer protocol to achieve high performance, and an increase in the number of users causes increasing latency.

It is popular for Local Area Networks (LAN).

7.3.6 Hybrid Methods 7.3.6.1 Hybrid - FDMA/TDMA FDMA/TDMA can be used as an alternative to the TDMA. The available wideband spectrum is divided into a number of subchannels that individually support a self- contained TDMA network.

FDMA/TDMA has similar attributes as TDMA.

The primary benefit of FDMA/TDMA is that it is often used in conjunction with Dynamic Bandwidth Resource Allocation (DBRA) to maximize spectral efficiency.

The primary drawback is greater complexity than TDMA.

7.3.6.2 Hybrid - FDMA/CDMA This technique can be used as an alternative to the CDMA. The available wideband spectrum is divided into a number of subspectra with smaller bandwidths. Each of

CMLP Study Page 68 August 2007 CMLP Final Report these smaller subchannels becomes a narrowband CDMA system having processing gain lower than the original CDMA system.

The benefits of this technique are that the required bandwidth need not be contigu- ous, and different users can be allotted different subspectrum bandwidths depending on their requirements. Also, there is superior mutual interference mitigation under conditions of net throughput large enough to drive the system into interference- limited operation.

The primary drawback is greater complexity than CDMA.

7.3.6.3 Hybrid – Time Division CDMA (TCDMA) TCDMA was initially proposed for cellular phone networks. In a TDMA system, different spreading codes are assigned to different cells. Within each cell, only one user per cell is allotted a particular time slot. Thus at any time, only one CDMA user is transmitting in each cell.

The primary benefit of TCDMA is avoiding the near-far effect since only one user transmits at a time within a cell. The primary drawback is greater complexity than CDMA.

Its typical application is for cellular phone networks.

7.3.6.4 Hybrid - Time Division Frequency Hopping (TDFH) Evolving from FHSS, TDFH has an advantage in severe multipath or when severe co-channel interference occurs. The user can hop to a new frequency at the start of a new TDMA frame, thus avoiding a severe fade or erasure event on a particular channel. TDFH combats multipath fading and avoids co-channel interference problems between neighboring cells if two interfering base station transmitters are made to transmit on different frequencies at different times.

The primary drawback is greater complexity than FHSS.

TDFH has been adopted for the GSM standard, where the hopping sequence is pre- defined and the user is allowed to hop only on certain frequencies which are assigned to a cell.

7.3.6.5 Time Hopping-Pulse Position Modulation (TH-PPM) Pulse position Modulation (PPM) is a form of signal modulation in which M message bits are encoded by transmitting a single pulse in one of 2M possible time-shifts.

CMLP Study Page 69 August 2007 CMLP Final Report

This is repeated every T seconds, such that the transmitted bit rate is M/T bits per second. TH-PPM is time hopping combined with pulse position modulation. Each user is assigned a unique time hopping spreading code. The spreading code is used to dither the transmitted pulse timing instants to achieve spread spectrum.

TH-PPM is a moderately mature technique. Its primary benefit is high spectral efficiency. The drawbacks include complexity due to the requirement of time synch and a sensitivity to multipath fading.

It is primarily useful for optical communications systems, where there tends to be little or no multipath interference.

7.4 List of Link Protocol Attributes

In this section, we focus on the functional attributes of the link layer. Recommenda- tions for specific standard link layer protocols is deferred, and will be coordinated with the SCaN Network Architecture Team (NAT) in order to incorporate trade analyses that consider the potential of similar functionality occurring at network layer or above.

7.4.1 Background: Data Link Layer concepts, services, and functions. Data Link layer is the second layer in the Open System Interconnection (OSI) reference model on communication systems. Conceptually, the OSI reference model maps out all the interdependency of various functions of communications system in a hierarchical, layered framework such that all functions associated with a lower layer will serve, directly or indirectly, the purposes of higher layer entities/processes. Thus, the OSI reference model can be understood from a top-down, service-oriented perspective, or a bottom-up, functional perspective. They are different ways of looking at the same thing.

To properly understand Data Link layer, we can explore its relationship to the physical layer below and the network layer above. The network layer is primarily concerned about routing data units (called packets) from the source node to the destination node via one or multiple relay (router) nodes. Although there is no explicit knowledge of the relationship or significance of individual packet to the original application level message, the network layer does have some end-to-end level information, e.g., source, destination, quality of service (QoS), etc. The packet size and routing/prioritization mechanisms are usually determined based on QoS needs. On the other hand, the physical layer is primarily concerned about optimizing the analog signal representation of each bit (or symbol) of data given the unique character-

CMLP Study Page 70 August 2007 CMLP Final Report istics of the particular wired or wireless communications medium so that the signals can be transmitted with as little distortion as possible and also as fast as possible. One can define the Data Link layer as a set of adaptation functions/procedures designed to match the communication capacity and characteristics of the physical layer with the service requirements of the network layer. The subset of data link functions/procedures that has to do with adaptation related to the network layer service requirement is called Logical Link Control (LLC) and those having to do with physical layer adaptation are collectively called Medium Access Control (MAC). The scope of a link protocol is one “hop”.

The link protocol attributes may be categorized as follows. We classify the functions into four broad classes, and each class then further identifies specific functions:

• Data Transfer – Transfer variable-sized service data units (SDUs) over serial links – Recognize data units and length at receiver – Provide Segmentation and Reassembly – Provide fill data when required by Physical Layer; synchronization – Provide Link Layer encapsulation and addressing – Provide compatibility with multiple network layer protocols (IPv4, IPv6, and legacy network layers) – Minimize overhead (impact on throughput/utilization) – Minimize impact on coding and lower layers

• Reliability and Quality of Service (QoS) – Support class of service capability at Link Layer • Prioritization – Provide strong error detection capability at Link Layer – Provide error correction via automatic retransmission – Support rich Link Layer metrics for accountability

• Channel Access and Usage – Operate over a shared channel • Virtual Channels • Medium Access Control (MAC) – Provide link establishment (hailing) – Provide channel management and link adaptation

• Link Layer Security In the following, we describe these link protocol attributes in more detail.

CMLP Study Page 71 August 2007 CMLP Final Report

7.4.2 Data Transfer: Transfer variable-sized service data units (SDUs) over serial links. The basic function of a link layer is to transfer data from one end to the other end of a link. These data units could be those defined by the CCSDS standards or network layer protocol units, such as IPv4 or IPv6 packets. The data units may be created by multiple independent applications, so the size of the data units may vary. The link protocol must therefore provide a way for the receive side to determine the start, stop, and length of each data unit.

7.4.3 Data Transfer: Provide Segmentation and Reassembly. Sometimes the network layer protocol data unit (N-PDU), which is also the service data unit offered to the data link layer (DL-SDU or SDU for short), may be too large for optimal processing and transmission across the link. One solution is to require the network layer to adapt to the link layer limitation by offering only SDU of certain size. Another solution is to implement segmentation and reassembly (SAR) capability in the link layer so that it can transform large SDU into data link layer protocol data unit (DL-PDU or PDU for short) of a different size. The segmentation occurs on the sender side; each SDU received from the network layer is broken into smaller piece of PDU with clear designation of its relationship to the original SDU. On the receiving side of the link the PDUs are reassembled to produce the original SDU.

Some physical layer functions such as coding handle data in fixed length blocks. In such cases, sophisticated ‘spanning’ techniques are utilized so that SDU(s) of arbitrary sizes can span into several fixed length, consecutive PDUs. The de-spanning processes where the SDU(s) are extracted out of the PDU(s) can be complicated. Of- ten SDUs that belong to separate “connections” cannot share the payload space in the same PDU frame due to different QoS requirements. Other reason for framing the data in fixed size has to do with optimizing channel access performance, e.g., fit- ting a single PDU into the time slot for an s-ALOHA channel.

The advantage of having link layer SAR functionality is that it simplifies the I/O process between the network layer and the link layer Service Access Point (L-SAP) because the link layer can receive large SDU; in other words, the network does not have to segment its data and “spoon” fed them to the link layer. However, the receiving side cannot deliver partial SDU to the peer network layer protocol, therefore requiring larger internal buffer space at the link layer when the SDU size is large.

SAR is a basic function common to a “convergence layer”, in this case between the network and data link layers. A wireless terrestrial example of link layer SAR is the Radio Link Control (RLC) of GPRS/EDGE. In the space context, the CCSDS Prox- imity-1 “Packet Delivery Service” provides the SAR function.

CMLP Study Page 72 August 2007 CMLP Final Report

Another example is the Licklider Transmission Protocol (LTP), which is an ARQ error control protocol that may be used at either the link layer or the transport layer. LTP includes SAR functionality to perform this function.

7.4.4 Data Transfer: Provide fill data when required by Physical Layer; synchronization Synchronization on the data link layer means the ability to find beginning and end of certain data segment within the stream of bits received by the physical layer. On the link layer there are in general four different levels of synchronization: (1) frame synchronization, (2) code block synchronization, (3) content synchronization, and (4) MAC synchronization.

Frame synchronization is usually achieved through the use of a special flag, or synchronization markers (a particular bit pattern with certain fixed length), inserted at the beginning of each frame so the receiving protocol can unambiguously identify the pattern with high probability. In general, longer markers are needed to assure high probability of detection and synchronization for operation under more challenging conditions. Code synchronization is used for the purpose of identifying edges of code blocks for decoding operation. Depending on the design of the system, code block synchronization may occur before, after, or simultaneously with frame synchronization. In general, code blocks and data frame do not have to align. However, one could combine code block and frame synchronization together to save overhead with some loss of flexibility in using variable frame length.

Content synchronization refers to finding SDU boundaries within the data based on the higher layer contents. For example, the edge between two network layer packets may occur inside the payload of the single frame. In order to separate the two packets, some information must be imbedded within the frame header or within the payload. To read the information from the header or the payload, the frame header must be identified first, which is the job of frame synchronization. So we see that in general, content synchronization requires frame synchronization, but not vice versa.

Lastly, there is MAC synchronization, which deals with the synchronization process for accessing a shared channel. There are a variety of issues but the most important one is on correct timing. If many nodes are sharing a single channel in time domain but their clocks are not properly synchronized, significant collision and performance degradation can occur.

CMLP Study Page 73 August 2007 CMLP Final Report

7.4.5 Data Transfer: Provide Link Layer encapsulation and addressing A link layer supports multiple data streams through virtual channels, but also supports data packets generated by different network protocols through encapsulation. The encapsulation function allows the link layer to recognize the protocol (e.g., IPv4, IPv6, Space Packet, etc) associated with the payload from a header field that contains the protocol identification address and to process the packet accordingly. One example would be the Ether Type indicator in an Ethernet frame.

In general, link layer addressing is not limited to the purpose of protocol identification, as in the case of providing encapsulation function. It can be used for general- purpose identification of the intended recipient(s) of a packet, be it a particular protocol, storage device, database, etc. Another form of addressing is for medium access control (MAC address) to identify the source and destination of a frame.

Encapsulation is also a basic convergence layer function that allows matching between systems such as Ethernet & SONET and adapting multiple packet traffic types to common underlying transport. A simple space example is the CCSDS Encapsula- tion Packet.

7.4.6 Data Transfer: Provide compatibility with multiple network layer protocols A link layer should not limit the choice of network layers that can be used above it. This is especially true, in the case of a service provider that wishes to provide a standard link layer interface to a wide variety of users.

7.4.7 Data Transfer: Minimize overhead (impact on throughput/utilization) It is understood that the functionality provided by the link layer comes at the cost of extra bits or bytes included in the link. The amount of this overhead should be minimized, such that the benefits of the link layer protocol outweigh the impact on data throughput.

7.4.8 Data Transfer: Minimize impact on coding and lower layers

The data link layer should be compatible with the underlying physical links by not requiring any performance requirements that are not within reason.

CMLP Study Page 74 August 2007 CMLP Final Report

7.4.9 Reliability and Quality of Service (QoS): Support class of service capability at Link Layer Quality-of-Service (QoS) can be specified in two ways:

 Quantitative performance metrics – QoS can be specified by quantitative performance metrics such as throughput, delay, delay jitter, packet loss rate, bit and frame error rate, and average and peak data rate. Usually three functions are required to support quantitative QoS metrics: acceptance, policing, and traffic shaping. The acceptance function receives QoS request from a user, looks at the current system resources, makes a decision on whether it can meet the service request or not, and if it can accept the service request, makes appropriate resource allocation and terms of the service agreement.

 The policing function is implemented on the link layer side of the interface, to monitor the flow of data and make sure it stays within negotiated parameters specified in the service agreement. The user cannot arbitrarily increase service demand on the link layer beyond what had been agreed upon. The traffic shaping function is implemented on the network side of the interface to make sure the statistical characteristics of the data stream (or the shape of the traffic) abides by the terms of the QoS agreement at the time of logical link crea- tion or MIB.

QoS can also be specified by functional features such as reliability, sequential delivery, etc. Here, the QoS metric is not quantitative, since a function is either present or not. QoS specified by features can be provided to user through properly activating the appropriate set of functions at the time of connection establishment.

Currently, the CCSDS standards only support QoS specification based on functional features. For example, in the Advanced Orbiting System (AOS) standard we find three ways of categorizing grades of service based on error control and sequence control:

a. Grade-3 Service – no error control, no sequence control

b. Grade-2 Service – FEC error control, data sequence preserved

c. Grade-1 Service – FEC and ARQ error control, data sequence preserved

CMLP Study Page 75 August 2007 CMLP Final Report

7.4.10 Reliability and Quality of Service (QoS): Provide strong error detection capability at Link Layer The network layer and applications above the link layer require at minimum, notifi- cation if any data received from the link layer may be in error. In the case of a system using automated data routing at the network layer, undetected errors passed to the network layer will cause inefficient bandwidth utilization.

7.4.11 Reliability and Quality of Service (QoS): Provide error correction via automatic retransmission Error control provides the mechanisms by which the imperfection of the communication system is made transparent to the higher layer. The basic means by which error control is accomplished are forward error correction (FEC), automatic retransmission request (ARQ), and hybrid schemes. RFC 3366 “Advice to link designers on link Automatic Repeat reQuest (ARQ)” is a summary informational document on link layer ARQ, and includes some discussion of its relationship with the higher layer Internet protocols.

The fundamental driver for choosing a specific error-control mechanism is the cost of retransmission. Sending retransmission request uses bandwidth; furthermore retransmitting data is usually not the most efficient way of correcting errors. For example, a 1kilobyte frame may have only 10 bit errors, which means 99.875% of the bits are received correctly. However, to correct that 10 error bits, we need to retransmit the entire frame. This means we have 99.875% overhead on our bandwidth usage when we retransmit a frame. In terms of delay, a retransmission request cannot arrive at the sender faster than the round-trip light time (RTLT). For a high delay-bandwidth product link such as the deep space link between a spacecraft and its Earth-based ground stations, the RTLT can be much longer than the transmission time of the entire message. This means the spacecraft or the ground station must maintain a connection waiting for indication of whether retransmission is needed, thus increase the delay by at least one way light time (OWLT) even when there are no errors. Despite its shortcomings, ARQ can provide 100% reliability where as FEC cannot provide such a guarantee. FEC also costs the system in terms of bandwidth and delay because it embeds redundancy in the data, thus lowering the effective data rate on the link, and requires sophisticated processing to recover the original data. One must explore the trade space thoroughly to find the most suitable error-control mechanism for any particular application.

The ARQ mechanisms are stop-and-wait (S&W), go-back-N (GBN), and Selective Repeat (or Reject) (SR). S&W is bandwidth inefficient unless the time-bandwidth product is very small (not typical in the space environment) or other traffic can be multiplexed while the process is waiting for acknowledgment. GBN may be used when the time-bandwidth product is reasonably bounded. Its complexity is simpler

CMLP Study Page 76 August 2007 CMLP Final Report

than that of SR, and was therefore chosen for e.g. the Proximity-1 protocol that is used in such scenarios as orbiter – surface asset communications. SR is the most bandwidth efficient, but requires greater buffer management. SR is part of the well- known HDLC standard. SR is part of the well-known HDLC standard as it may be used in the Asynchronous Balanced Mode (ABM).

The Licklider Transmission Protocol (LTP, see http://tools.ietf.org/group/irtf/draft-irtf-dtnrg-ltp-07.txt) uses a SR mechanism but is designed specifically to accommodate the extremely long signal propagation de- lays, frequent and lengthy interruptions in connectivity, low levels of traffic coupled with high rates of transmission error, and meager bandwidth and highly asymmet- rical data rates. It is highly stateful in order to avoid negotiation exchanges that are inappropriate in deep space links. It assumes that it receives cues from the underlying MAC sublayer or onboard process (knowledgeable of when the link conditions will be right). The timers use “punctuated” time, i.e. they only decrement when data could be arriving over the link.

In general, a combination of FEC and ARQ will provide good performance under a variety of situations. Typically one would use FEC to bring the error performance up to certain level but take care of the residual errors by ARQ scheme. While Hybrid ARQ (HARQ) is not currently in use in space applications today, it is being applied to terrestrial wireless applications and may find use in space. Such Incremental Re- dundancy (IR) techniques consist of the source initially transmitting the frame with sufficient parity to determine whether it is received correctly. If not, then rather simply retransmitting it anew, the receiver maintains its received copy, and the transmitter sends additional parity that the receiver may combine with the previously received frame.

Sequence control means ordered delivery of SDUs. Typically, a link connection is mapped to one physical connection; therefore, ordered delivery is assumed except when an ARQ mechanism such as selective repeat is used. There is a possibility that a SDU transmitted later may arrive completely and correctly and become available for delivery before another SDU that were sent earlier but are still waiting for retransmission to correct errors. In such case, sequence control may be implemented to buffer the SDUs temporarily and deliver them in the same order they are given by the network layer. For systems using go-back-N or stop-n-wait ARQ, the error control mechanism automatically enforces sequence control on the received data.

7.4.12 Reliability and Quality of Service (QoS): Support rich Link Layer metrics for accountability The provision of link layer metrics is critical for troubleshooting complex links. By providing data beyond the physical bit error rate measurements, link layer metrics

CMLP Study Page 77 August 2007 CMLP Final Report

allow the isolation of errors particular to specific user or system. In non- troubleshooting scenarios, the link layer metrics allow the bandwidth usage to be monitored based on user or application.

7.4.13 Channel Access and Usage: Operate over a shared channel: Virtual Channels Virtual channels provide the means by which multiple traffic streams can share a single link layer connection. A virtual channel provides a simple framework by which QoS and resources can be differentiated and allocated to individual data streams.

One function that naturally arises from having multiple virtual channels is prioritization based on QoS. A data stream having a higher QoS requirements could be given higher priority to buffer space and access to the physical channel for reduced packet loss rate and latency. Prioritization mechanisms are most naturally implemented by queuing policies and multiplexing rules. Many queuing policies implemented on today’s Internet routers, such as strict-priority, weighted round robin, and weighted fair, could be applicable for use over space links.

The multiplexing of several network layer connections onto one virtual logical connection of the LLC layer may be considered a convergence protocol function. Examples of link-layer Virtual Channels in space applications are CCSDS AOS and CCSDS Proximity-1 “I/O ports”.

7.4.14 Channel Access and Usage: Operate over a shared channel: Medium Access Control (MAC) Medium access control has to do with a whole class of functions that enables the efficient sharing of a single communications medium by multiple entities. In fact, this function is so essential and fundamental in wireless communication that the IEEE 802 standards body adopted it as one of two sub-layers within the link layer. Classic MAC functions can be categorized into two different approaches: (1) scheduled (contention-free) and (2) random access (contention-based). Contention-free MAC technique typically divides communications resources in the time, frequency, or code domain in such a way that completely avoid collisions. Each node can have either fixed allocation, such as time division multiple access (TDMA), or demand- driven allocation using procedures such as polling or token passing. The advantage of this approach is that resource allocation is controlled, and latency is predictable; however, efficiency could be low when the traffic loading on the system is more dynamic. Contention-based MAC system achieves higher performance by allowing controlled collision in order to improve system efficiency under dynamic traffic conditions. Currently, space link operations have treated each link as a point-to- point link, which is true most of the time due to the use of directional antenna and

CMLP Study Page 78 August 2007 CMLP Final Report

the fact that there are really very few entities spatially. Even in cases where there is the possibility of multi-access, the ground operator would schedule long (e.g., seconds, minutes) disjoint communication periods instead of allowing intermixing of data streams from different assets over the same transmission medium. However, if one can envision a future where the number of spacecrafts is sufficiently large to create a rich networking environment, Medium Access Control (MAC) will be crucial in providing a distributed, channel sharing mechanism at finer time resolution for a large population of spacecraft and relief the burden on Earth-based space link planning and scheduling.

Medium Access Control is active procedures taken by nodes to share a common communications resource; more generally, Multiple Access may be achieved by pas- sive means, such as simple pre-allocation of the resource according to some partition. Multiple Access may be considered as a subset of the Data Link Layer (and hence “Link Protocols”) although there is overlap with the underlying coding and physical layers as well. The multiple access attributes are addressed in a separate document provided by the CMLP Multiple Access Subteam that has been specifically identified; the reader is directed to that document for that information.

It is perhaps worthwhile to point out the special case of MAC when there only two users, i.e. a point-to-point link. This is referred to as Half-Duplex (HDX) Rx/Tx turnaround control. In space systems, it is common to use full duplex (FDX) channels, partly because this permits coherent RF turnaround lock that may be used for precise ranging (navigation). However, studies have demonstrated that in orbiter – lander relay applications, the asymmetry of the traffic flows generally favors a HDX link, in addition to its ability to dynamically accommodate bandwidth in either direction. An issue of concern is the lifetime of the hardware Rx/Tx switch, which may influence the choice of turnaround times. An example space link protocol supporting HDX (as well as FDX) control is CCSDS Proximity-1.

The related function of flow control is mentioned here. Flow control manages the amount of data flow across the link to prevent congesting the receiver’s storage buffer, as well as to enforce QoS requirement negotiated for a connection. The link layer protocol controls the data transfer rate available to the sender at the L-SAP. Open- loop flow control mechanism monitors and controls the sender’s average and peak rates based on constraints established at the time of connection establishment or by the Management Information Base (MIB); closed-loop flow control requires the sender to also take action based on the presence or absence of feedback information from the receiver. For example, the sender may receive a direct request from the receiver to throttle down on the rate of transmission or voluntarily reduce data flow when there is a prolonged absence of positive acknowledgements from the receiver.

The LTP protocol does have congestion control mechanisms identified.

CMLP Study Page 79 August 2007 CMLP Final Report

7.4.15 Channel Access and Usage: Provide link establishment (hailing) A link layer connection is logical, thus it is transparent to the underlying physical layer. Two peer entities can maintain a logical link connection even when the physical link is temporarily absent. Therefore it is conceivable for a logical connection to span across multiple time-disjoint physical layer connections. Communications can suspend when the physical layer is absent and resume from the same place when the physical connection is restored.

A connection is a stronger notion than just having information about each other; a connection means that the shared state information between two end-points of a link will affect their individual behaviors resulting in coordinated operation. To negotiate, configure, establish and maintain a connection involves two-way communications such as a hand-shaking process, where the two end-points exchange and update state information and negotiate how they will operate with each other. The benefit of having a connection is the efficient usage of resources through coordination. If A wants to send data to B, it makes sense to make sure that (1) B is available and ready to receive the data and (2) the appropriate level of resource is dedicated to support the connection. When the communication associated with a connection is complete, the resource dedicated to the connection can be explicitly released for reuse for other connections. Thus inherent in the concept of connection is resource management.

An example of connection management in a space protocol is the CCSDS Proximity- 1 establishment and disestablishment of a “session” by means of the “hailing” procedure. Such a process configures the (initial) parameters to be used for the link. If automatic repeat request (ARQ) error control is used, the accounting for this is scoped within only a single session. Links that are characterized by large bandwidth-delay products (such as typical interplanetary links) may not be able to continuously support the two-way establishment of connections, and a connectionless data link will be preferred. In such cases, higher-layer protocols might provide connection services.

The Link Establishment process cuts across more than just the data link protocol, since in general it may determine such aspects as data rate and frequency band to be used. Demand Access is the capability in which the link may be established with little or no prior planned sequence management. Space links are typically pre-scheduled according to sequence commands previously uploaded into the spacecraft (and similarly the ground station is scheduled when it is an end point) so that the many op- erationally competing elements may be taken into account, such as power that

CMLP Study Page 80 August 2007 CMLP Final Report

would be otherwise used by other spacecraft subsystems, orientation of the vehicle, etc. Nevertheless, Demand Access will be fruitful for cases of emergency needs, for opportunistic science, and may be managed within certain operational bounds of spacecraft system nominal operations.

Associated with the hailing (connection startup) process is the node’s energy management, i.e. whether it may be in radio sleep mode or not. While this function operates beyond the confines of pure “data link protocol” it certainly uses these protocols for these procedures.

7.4.16 Channel Access and Usage: Provide channel management and link adaptation Link Adaptation techniques dynamically adapt transmission rate, modulation, code rate, slots/packet (for TDMA systems), packet size, preamble length, power, and other real-time operational reconfiguration that attempt to adapt to the dynamic condition on the channel as well as changing traffic conditions. This requires fast feedback channel state information (a side information channel). One could make a distinction between the control functions for QoS support and channel adaptation. The control algorithm for QoS support deals with providing each individual logical link the required QoS; the algorithm for adaptive link control is aimed at maximizing the aggregate performance of the link. In IEEE 802.16, the QoS support is provided on the aggregate level (per service station, which may have multiple logical connection to the base station) and as well as on a per logical connection basis; while link adaptation is only performed on the aggregate level.

An example terrestrial system is the GPRS/EDGE system. In space systems, the technique is not well documented in unclassified domains. However, there is a plan to upgrade the radio on MRO that is using the Proximity-1 protocol to enable an adaptive data rate capability.

7.4.17 Link Layer Security Link security may be deployed in conjunction or as an alternative to application, transport, and network layer security measures to protect information content. Secu- rity may also be deployed as a measure of protecting communication resources from unauthorized access that reduces the availability of bandwidth and processing resource for legitimate users.

7.4.18 Link Layer Functionality by Link Type In this final link layer subsection we do not provide an additional link protocol attribute but rather consider the previously defined functions in terms of three catego- ries of space links: long haul point-to-point, short range (or proximity), and sur-

CMLP Study Page 81 August 2007 CMLP Final Report

face/local area network links. These link types are defined for the purpose of distinguishing link protocol functionality and differ from those defined in Section 5.

7.4.18.1 Long Haul Point-to-Point Link The main characteristics of a long haul point-to-point link are the long propagation delay and large signal strength attenuation. Therefore, ‘chatty’ communications could be prohibitively inefficient. For example stretching TCP or other such connection-oriented protocols over challenged long-haul links, even when feasible, carries a significant overhead. Similarly an interactive web-based application is difficult to support in such environment. The primary functionality of a long-haul point-to- point link is trunk line communication providing reach-back capability for remote spacecraft/astronaut. To improve efficiency one expects that multiple traffic streams be multiplexed onto the same trunk line. Availability of such a link is usually highly predictable, as it is dominated by ephemeredes, so the need for dynamic management is not a significant design driver. However, such a link will be carrying a variety of data types for multiple users. These include science telemetry, critical operational data and command, public out-reach video, and other less important house- keeping data types. A strong error correction/detection capability will be necessary. Virtual channel support is critical to provide a mechanism for traffic separation and bandwidth management. Data prioritization is important for ensuring that critical and real-time data are delivered with the lowest possible latency.

7.4.18.2 Short Range Proximity Link Proximity link covers those situations where the propagation distance is not so large to prohibit the use of (1) a “hand-shaking” process for setting up connection- oriented services (such as Proximity-1’s concept of a link layer “session”) and (2) wider beam-width or even omni-directional antenna. We can divide the application domain of proximity link into several areas distinguished by the traffic characteristics:

a. Relay link for remote assets – These are for example the point-to-point link between an asset on the surface of Mars and an orbiter. It can also include an orbiter-to-orbiter link. These links are characterized by short and infrequent contact durations, and are intended to carry large volume of science and engineering data during the contact period. Communication windows are almost always manually scheduled as connectivity is again mostly driven by ephemeredes.

Examples of such links include the MGS and Odyssey science orbiters on Mars that provide data relay services to the MER rovers. [MGS has recently been lost.] Both MGS and Odyssey have their own science objectives; therefore, their design and operations are not optimized for relay performance.

CMLP Study Page 82 August 2007 CMLP Final Report

Their orbits are primarily chosen for science reasons, not relay coverage, so the communications windows with surface assets are typically short and infrequent. These types of relay links provide a significant link budget advantage to the surface assets by leveraging the long-haul communication capability of the orbiting assets. The primary functional requirements are connection management and error control. A moderate level of topology dynamics may arise when multiple surface assets are simultaneously present within the orbiter’s antenna coverage. Medium access control function may be required to coordinate sharing of uplink to the orbiter. It is also possible that the use of an omni antenna and a low altitude orbit creates a more challenging channel environment. Therefore, rate adaptation techniques are expected to enhance performance.

b. Near Earth and Cislunar Communications – The primary difference of near Earth relay and deep space planetary relay is the potential requirement to support manned missions. Real-time voice & video for astronauts, command & control, Internet access, email, and a variety of other interactive applications changes the characteristics of the traffic significantly. The AOS standard is designed to support isochronous traffic, and is currently used for the Space Shuttle and the International Space Station. Rendezvous and docking operations will raise the requirements for guaranteed bandwidth and low latencies.

c. Inter-spacecraft communication for formation flyer and constellation – Preci- sion Formation Flying (PFF) will enable precise scientific observation and measurement of physical phenomena for space science. Missions such as Ter- restrial Planet Finder (TPF) need link layer communication protocols capable of meeting the latency, bandwidth, and reliability requirements for distributed navigation and formation control loops. Some of the missions involve spacecraft operating in close range at high levels of network self-organization, which makes the use of omni-directional antennae both feasible and important for network initialization and formation transition. The required functions include a MAC function capable of supporting the entire span of multi- access control, from asynchronous random access to contention-free communications.

d. Relay Orbiters – The deployment of a dedicated communications orbiter in high altitude orbit would introduce a new range of link layer functional and performance requirements. A high altitude orbiter can provide long passes, several hours a day, compared to current coverage of approximately two passes per day, each pass lasting only 2 to 8 minutes. The data rate would be pushed from 128kbps or 256kbps (for Odyssey) to 8Mbps or higher. The bandwidth-delay product would be raised to a point where the current standard such as the Proximity-1 protocol would need certain performance fine-

CMLP Study Page 83 August 2007 CMLP Final Report

tuning and also functional extension to work. For example, the upper bound on the half-duplex transmission windows size for a such a high orbiter operating at 8Mbps can go as high as 200 or even 2000, which exceeds the current limitation on the frame sequence number in Proximity-1. In addition, a high orbiter that can provide relay service to multiple surface and in-orbit assets calls for demand-driven MAC functionality to be implemented. Future manned missions to the Moon and Mars may required the capability for un- scheduled, dynamic sharing of uplinks and downlinks between surface assets and a relay orbiter such as the Lunar Relay Satellite (LRS) envisioned to communicate with multiple assets on the lunar surface including rovers, landers, EVA astronauts, habitat modules and other mobile and non-mobile equipment.

Channel adaptive techniques such as coded modulation will play a significant role in pushing the data volume up. Automated data rate changes would be facilitated by the appropriate exchanges of link layer directives and radiometric information. Also the buffer management function faces new challenges since the amount of data that can be transmitted during a single pass is no longer predictable – a storage management function needs to be built into the link layer and working closely with on-board resource manager.

e. Entry, descent and landing (EDL) link – The communications on EDL are brief but highly critical. The nature of the event makes use of relay assets ideal for capturing EDL data. Techniques to ensure robust performance are highly sought. Similarly, coverage of ascent vehicle events (e.g. Mars Ascent Vehicle in a sample return mission) require high reliability. It is noted that multiple relay assets are preferred in EDL conditions for purposes of positioning/navigation (to create a triangulation baseline). The relay of a guidance solution to a simple probe being targeted to a precise location in which the link is challenged by plasma effects might be aided by a link protocol.

7.4.18.3 Surface/Local Area Network Links a. A wireless local area network (WLAN) infrastructure will connect astronauts performing EVA and rovers with landers and life support systems. It provides an alternative to wired cables with savings on mass and volume. WLAN should have the characteristics of being easy to setup and share, so dynamic connection management and MAC are critical for such links.

b. Surface-to-surface relay link i. Surface repeater – a special infrastructure node such as a repeater tower can be set up on strategic locations of concentrated exploration activities. Such a link is expected to operate at multiple rates and offer a variety of

CMLP Study Page 84 August 2007 CMLP Final Report

error control methods such as ARQ and FEC in order to provide the desired delivery fidelity with both nearby and distant users who radio may have different capabilities. Dynamic connection management is critical as mobile users may roam in and out of the coverage area of the repeater and exchanging interactive messages. ii. Surface mesh network links – when there is a sufficient presence of surface assets, a multi-hop mesh network can provide relay between any two surface assets. Such a network consists of repeater links, WLAN links, and in some cases even the orbiter relay links in order to maintain connection between two distant users. Again, dynamic link management is critical.

c. Surface-to-orbiter link – this type of link provides beyond-the-horizon range extension for distant surface assets. Onboard store-and-forward capability will enhance the range extension capability by allowing assets separated by a distance greater than the diameter of the relay orbiter’s antenna foot print to communicate.

CMLP Study Page 85 August 2007 CMLP Final Report

8 Figures of Merit (FOMs)

The figures of merit (FOMs) are defined as follows:

a. Supports legacy missions (time span, percent of features) This is more than a statement about supporting missions that are currently flying. At any point in time, this FOM is still important. For example, in 2025 we will still need to support missions launched in 2020. Hence, we view this FOM as a sliding window of support. The size of the window defines the metric, along with the number of legacy features that are supported.

b. Spectrum utilization This FOM reflects spectrum constraints, such as channel bandwidth (Hz) and Power Flux Density (PFD) limits (W/m2), and bandwidth efficiency (bits/sec/Hz). It includes the use of algorithms to prioritize data to get the best use out of the available spectrum.

c. Power efficiency (PT/(RN0) to get required performance) Here PT is the total power (the sum of modulated data power Pd and residual carrier power Pc, if any), R is the data rate, and N0 is the noise spectral density. Note that PT/(RN0) = (Pd+Pc)/(RN0) = Eb/N0 + Pc/(RN0). Thus, for suppressed carrier modulations (i.e., Pc=0), PT/(RN0) is identical to the familiar power efficiency measure Eb/N0, while for residual carrier modulations (i.e., Pc>0), PT/(RN0) correctly accounts for the additional power used in the carrier. We will also allow for prioritization schemes as discussed in “spectrum efficiency.

d. Infrastructure burden (Percent cost increase) This is similar to “user burden” but the Space Communications and Na- vigation (SCAN) infrastructure elements incur these costs. It is critical to measure the required cost increases on both sides in order to make programmatic decisions on investment.

e. Alignment with international standards (Probability of alignment) Alignment is very difficult to measure. We decided to measure this as a subjective probability that the recommended scheme will become an international standard. If, for example, all aspects are already standards, then this FOM is a 1. If there are elements that we believe can never become standards, it is a 0. International standards includes those managed by CCSDS, ITU/SFCG, IOAG, and IEEE.

f. Provide radiometrics for navigation (Accommodation % cost increase)

CMLP Study Page 86 August 2007 CMLP Final Report

Radiometrics may be required on various links to provide navigation or radio science. We will assume appropriate requirements for these dependant on the link. The data types being considered will include, Dop- pler, Doppler rate, ranging, and differential techniques, as appropriate. We will measure the additional cost (as a percent) to accommodate the required data types at the required performance levels.

g. Robustness This FOM considers robustness to short term signal disruptions, whether caused by mechanical failure (e.g. inadvertently spinning spacecraft) or by sensitivity to electronic signal distortions. This FOM also considers the effects of interference, including electromagnetic interference (EMI), inter- symbol Interference (ISI), and cross-channel interference (CSI). We will measure the additional power required to mitigate these sources to the levels required by the individual link classes.

h. Latency (seconds) Latency will come from at least three major sources: end-to-end latency, time to acquire signals, and latency from signal processing. These are somewhat independent. The sum of these will be used for this FOM.

i. Technology maturity (TRL) We will estimate the technology readiness level (TRL) of the critical components of the system and make a judgment as to the overall maturity.

j. Capacity We will consider the aggregate capacity of simultaneous links between multiple elements, including the data rates that can be supported between individual elements and the number of simultaneous links.

CMLP Study Page 87 August 2007 CMLP Final Report

9 Navigation Considerations

This section provides an overview of radiometrics as needed for navigation and ana- lyzes the effects of signal design on radiometric performance. It shows that Doppler performance is not a significant signal design discriminator and estimates ranging performance for various ranging signals.

9.1 Navigation within the Space Communication Architecture

NASA’s future Navigation and Time architecture is outlined in its documented recommendations for the 2005-2030 Space Communication Architecture (SCA).1 The primary navigation function is trajectory determination, which interacts with vehicle guidance in a cyclic sequence of navigation observations, trajectory updates, and guidance maneuvers as a spacecraft progresses toward a destination, whether in free space or on a planetary surface. SCA navigation encompasses elements of both radiometric and autonomous services.

Radiometric services include those whose information is derived from “navigation- only” transmissions (e.g., sequential ranging tones as used in the DSN 26-m subnet, regenerative PN ranging as implemented on the New Horizons mission to Pluto,iii or the Global Positioning System, GPS) as well as from direct sequence PRN codes modulo-2 added to the data transmitted by TDRS and regenerated and returned by the user to provide both spectrum spreading and ranging services.

Autonomous services are those for which the user makes observations of naturally occurring phenomena, including: planet and star sensing, landmark tracking, gyro or accelerometer readings, wheel odometer cumulatives, and videocam frames.

Navigation requirements that fall outside the SCA scope, in the sense that they are user-supplied as needed, include radiometric and radar-based aids for rendezvous and docking.

Throughout the history of spaceflight NASA has provided primary navigation and vehicle guidance for spacecraft in widely varying locations, ranging from near-Earth missions to interplanetary flights. Earth-based radiometric tracking lies at the heart of most of these operations and will continue its key role in spacecraft navigation. Although GPS has a radiometric basis, it can provide an autonomous capability to a spacecraft using an onboard receiver to estimate—continuously or sporadically—the spacecraft state vector (position and velocity) and communicate the results to ground, as opposed to full reliance on ground-based tracking. Away from Earth the GPS may be replaced or supplemented by tracking aids such as planetary surface

CMLP Study Page 88 August 2007 CMLP Final Report beacons or in-situ navigation capabilities (e.g., relay satellite constellations) em- placed near planetary surfaces.

9.2 Requirements and Methods

In developing the SCA, the Space Communication Architecture Working Group (SCAWG) summarized key navigation performance requirements, position and velocity, for a number of capabilities NASA will require of future missions. These are summarized in Table 9-1.

CMLP Study Page 89 August 2007 CMLP Final Report

Capability Needed (3-sigma) 3-D Position 3-D Velocity Surface Operations 30 m TBD Global Surface Operations 30 m TBD 10 m (1-sigma) reconstructed (TBR); Relay Spacecraft TBD 100 m (1-sigma) predicted (TBR) Non-precision Landings 5 km @ landing 1 km (unaided) and LSAM Landings 100 m (aided) Precision Landings 100 m @ landing* 50 m with short la- RLEP Landing tency (post processing) Surface Rendezvous 10 m @ landing TBD Ascent (surface location) 10 m @ liftoff \Not a Driver Rendezvous (@ relative navigation 10 cm / s (rela- 500 m (relative) initiation) tive) Docking and Berthing (assuming iner- 1 km 50 cm / s tial navigation available as backup) In-space Servicing (assuming inertial 1 km 50 cm / s navigation available as backup) Constellations* 100 m (absolute) 10 cm / s Formation Flying*—coarse 10 m (relative) 3 cm / s Formation Flying*—precision 3 m (relative) 3 mm / s Formation Flying*—very high preci- 3 cm (relative) 0.03 mm / s sion Libration Point Stationkeeping 50 km 2 cm / s Fly-bys, Impulsive Transits TBD TBD Fly-bys, Low-thrust Transits TBD TBD * A constellation, i.e. a cluster of spacecraft, requires absolute position and velocity knowledge with respect to Earth. Constellations that require formation flying, i.e. maintaining precise offsets between spacecraft, have requirements for position and velocity relative to each other. * The LAT2 study targeted 1 m accuracy for precision landing Table 9-1 Key Navigation Performance Requirements

After studying the requirements and the assumed future capabilities, the SCAWG was able to make a tentative allocation of appropriate navigation data types according to mission phase, as shown in Table 9-2. The underlying logic comes from numerous point studies carried out by SCAWG team members.

CMLP Study Page 90 August 2007 CMLP Final Report

Mission Phases Navigation Data Types Launch/Ascent supported by angles-only tracking Launch/Ascent/ GPS (Space-Based Range using GPS and TDRSS) Entry/LEO Entry/LEO supported by Earth-based range/Doppler and GPS pseudo-range High Earth Orbit Earth-based range/Doppler Navigation GPS pseudo-range and data message Deep Space Na- Earth-based range/Doppler and Very Long Baseline Interfer- vigation ometry (VLBI) data types Earth-based range/Doppler can meet needs for all orbiting users Lunar-orbiting range/Doppler can meet needs for other or- Lunar Vicinity biting users and is required if mission critical descent opera- Navigation tions out of Earth view are to be covered Lunar-orbiting range/Doppler is adequate for surface users given a certain latency with user burden constraints Earth-based range/Doppler can meet needs for all orbiting users Mars-orbiting range/Doppler can meet needs for other orbiting users Mars Vicinity Mars-orbiting range/Doppler is adequate for surface users given a certain latency with user burden constraints Mars-orbiting range/Doppler is required for precision approach and landing Table 9-2 Navigation Data Types and Mission Phases

The role of GPS in NASA’s future has been the object of much study. While the SCAWG recognized that many missions can satisfy their navigation requirements using radiometric capabilities of the SCA, missions also may decide to use GPS, either in addition to or in place of radiometric tracking. GPS is the primary SCA alternative for space vehicle navigation in Earth vicinity as far as geosynchronous orbit. Additionally, GPS may prove useful for launch and re-entry phases.

GPS signals may be utilized in specialized applications with apogee above geosta- tionary altitude and perigee altitude of 8000 km or less. These include applications whose performance accuracies exceed those obtainable through radiometric tracking, and those using GPS-precision timing onboard the vehicle for navigation or; or scientific measurement purposes. Navigation beyond GEO requires application- specific GPS receiver architectures, key features of which include: enhanced acquisi-

CMLP Study Page 91 August 2007 CMLP Final Report

tion and tracking algorithms; integrated, extended Kalman filters, onboard clock models, and ultra-stable oscillators.

9.2.1 Analysis of the Navigation Requirements Radiometric measurements from which navigation information can be estimated consist mainly of Doppler and range data. Position determination is normally associated with range, and velocity with Doppler, but in fact both may contribute to either estimate. The ability to extract each of these from a signal poses requirements that are discussed individually below.

9.2.1.1 Range The most stringent range requirement found in Table 9-1 for typical NASA operations is 10 m, 1, for landing, liftoff, and coarse-resolution formation flying.2 To meet this is the equivalent of measuring the arrival time of a signal to 33 ns. In actu- ality, 33 ns is the result of processing the several measurements through a position estimate calculation for which the individual measurement errors are attenuated by a geometric dilution of precision (GDOP) factor. Depending on the ambient geome- try of the sensors and users, the component measurements typically can be somewhat more erroneous than the net position.

A useful ranging signal, or set of signals, has the properties that: (1) each signal is uniquely identifiable; and (2) its time of arrival of each can be measured to within a specified accuracy. Uniqueness may be established in one of two ways. If a signal is the only one that can occur in certain frequency band and region of space, existence and uniqueness are close to one and the same. In a multiple access situation, signals must differently exploit the available degrees of freedom in time, frequency, and space to be distinguishable.

Orthogonal signals are distinguishable and non-interfering, but difficult to maintain in some environments in which coordination of transmissions is impossible or unde- sirable. Statistically orthogonal signals such as are typically used in CDMA systems enable multiple access to a channel with a small, controllable amount of mutual interference.

The ranging potential of a signal may be assessed from a quantity called the Cramér- Rao bound, which is a lower bound to the standard deviation of arrival time for un- biased estimates on an additive, white, Gaussian noise channel. No estimate will

2 There is a 3m, 1 requirement for formation flying that might fall outside the infrastructure-provided services.

CMLP Study Page 92 August 2007 CMLP Final Report

achieve a smaller standard deviation, but it is not assured that any estimate will actually achieve Cramér-Rao bound performance.3 The bound takes the formiv 1 ( ˆ)  , (9-1) BW /2 NE 0intrad-rms

where ˆ is the arrival time estimate, / NE 0int is the integrated signal-to-noise ratio in the received signal, and BW rad-rms is the root-mean-squared bandwidth of the signal in rad/s, given by

 2/1  2    2 )( dS     BW   rad-rms    , (9-2)  )( 2 dS         Equation (9-1) is often called the ranging equation, because the range error (in meters) corresponding to (9-1) is c (ˆ) , where c is the speed of light in a vacuum, approximately 3 10 8 m/s. The RMS bandwidth shares characteristic properties with other bandwidth definitions such as scaling in time and frequency. If signal ts )( has RMS bandwidth 1 , then Tts )/( has corresponding bandwidth 1T . In (9-2), there is a mathematical analogy between RMS bandwidth and the standard deviation of a probability density, where the power spectral density (PSD) of the waveform plays the role of the probability density function—which is normalized to integrate to 1—and  plays the role of the random variable. Under the assumption of zero mean the formulas are formally identical. Both observations confirm that wideband character is necessary, if not sufficient, for a signal to have a large RMS bandwidth and small arrival time estimation error. Equivalently, if the signal is wideband, it will have a deterministic autocorrelation function with a narrow central lobe of width inversely proportional to the bandwidth. Optimum receivers for estimating time of arrival usually form the autocorrelation function by correlating the received signal with a local replica and choosing as the arrival time the time at which the computed correlation function is maximum. A narrow central lobe of the autocorrelation function assures accurate arrival time estimation if the a priori uncertainty in the arrival time is sufficiently small, i.e., on the order of the autocorrelation mainlobe width. The property is not sufficient to assure accurate arrival time estimation in general, however. If there are sidelobes of competing amplitude there is the potential for large, ambiguous errors, meaning er-

3 Signals whose received signal-to-noise ratio, 2 int / NE 0 , falls in to the reliably detectable range, e.g.,  10 dB, will typically achieve ranging performance near the C-R bound.

CMLP Study Page 93 August 2007 CMLP Final Report

rors whose values are near the times at which the various sidelobe peaks occur. Good ranging signal design not only takes advantage of any large, available bandwidth, but also eliminates high sidelobes that can compete with the mainlobe in noi- sy observations to generate estimates with large ambiguity errors. Ambiguity resolution need not be entirely built into one signal. Tone ranging systems use a succession of signals at different offsets from the carrier to achieve ambiguity resolution sequentially. No sine wave has a “good” autocorrelation function in the sense defined above, but a composite set of tones can yield both fine resolution and freedom from ambiguous errors. According to (9-1), estimation accuracy improves ( decreases) linearly with increasing bandwidth but as the square root of increasing SNR, showing why bandwidth and its efficient use is important to ranging in almost any scenario. Ranging signals may be used in a one-way or two-way mode. The one-way para- digm requires that accurate time be available at each terminal. In two-way ranging, the received signal is turned around at an ideally fixed and known delay, and the relative time knowledge at the two terminals drops out of the equation. There are two methods of turn-around ranging. In one case the intermediate receiver may function as a transponder that simply filters the input signal to it bandwidth, translates it in frequency, and retransmits to the originating station. In a second method, regenerative ranging, the two-way path is broken into two separate links by placing a full receiver for the initially transmitted code at the intermediate terminal, usually the spaceborne end. The incoming code is acquired and tracked, and a replica code is locally generates in synchronism with the tracked code—or at a fixed or measured offset—and retransmitted. If the offset is measured, its value is included in the returned data. The regenerative method, although long known, has only recently been introduced into the NASA repertoire of techniques. Regenerative ranging can improve the link budgets requirement considerably. Be- cause the receiver acquires and tracks the code, the residual noise is restricted to the bandwidth of the delay-lock loop, which can be in the few Hz region, or below. In contrast, a transponder retransmits receiver noise across the full code bandwidth. The difference can amount to many tens of dBs in the regenerated output. Regeneration also influences the selection of ranging signals. A spacecraft transponder is indifferent as to the specific code it receives, but the code structure is more a matter of concern since the remote terminal must track the code. For this purpose NASA has developed codes that intentionally do not take on the appearance of pseudorandom sequences, but are instead specialized to the demands imposed by placing a range code receiver in deep space, primarily simplifying code acquisition and tracking. These alternatives are composed of logical combinations of several short, periodic codes, put together in a way that accomplishes two distinct goals: (1) the code is

CMLP Study Page 94 August 2007 CMLP Final Report

simpler to acquire than a PN code because the component codes can be acquired individually, and from their epochs the overall epoch can be found; and (2) the resulting code is a square wave with occasional phase reversal that provide the ambiguity resolution function.v (An alternative viewpoint is that ambiguity resolution is inherent in the component code structure). This “almost square-wave” structure places most of the code power into the fundamental frequency, enhancing the attainable fine resolution. Regenerative ranging has been implemented in the New Horizons spacecraft launched 19 January 2006 on a mission to Pluto and the Kuiper Belt via Jupiter, which it encountered during the first half of 2007. Arrival at Pluto is scheduled for July 2015.

9.2.1.2 Doppler Doppler shift can be used to measure the component of velocity along the line con- necting a transmitter and receiver and is a tracking aid long in use. To measure Doppler requires that either a residual carrier component be present in the received signal, or a suppressed carrier (or a carrier harmonic) is reconstructed from which frequency estimation may be accomplished. Common modes of Doppler treatment are known as one-way, two-way, and three- way. In one-way tracking, the spacecraft generates a frequency that is received and measured for Doppler shift at a ground station by comparison to a local reference standard. For two-way, the signal originates at the ground and receives a phase- coherent turnaround at the user spacecraft, resulting in a signal whose frequency can be compared to the transmit frequency at the ground station. Three-way tracking results when a second ground station receives a coherently turned around signal in response to a two-way tracking uplink from a first ground station. Doppler pre- compensation is sometimes used in the attempt to preset the frequency of a received signal and keep the range of arrival frequencies small. Most of the modulation types considered in this CMLP study call for coherent reception and are therefore consistent with extracting Doppler measurements. All forms of phase modulation wherein a data symbol is mapped into a specific phase relative to the carrier are coherent. These include BPSK and a number of QPSK variants. Differential phase encoding, in which the phase of a new symbol is referenced to the phase of the previous one, does not require phase tracking, but only frequency tracking and stability over adjacent symbols for correct demodulation. FSK may be demodulated noncoherently. The receiver for a modulation method that does not require phase-coherent tracking for demodulation may track phase for purposes of frequency measurement. The Cramér-Rao bound for cyclic frequency estimation is

CMLP Study Page 95 August 2007 CMLP Final Report

1  ( fˆ)  (9-3)  rms /22 NET 0int ˆ where f is the frequency estimate, / NE 0int is the integrated signal-to-noise ratio in the received signal, and Trms is the root-mean-squared duration of the signal, ts )( , given by 2/1    2 2    0 )()( dttstt    T   rms    , (9-4)  )( 2 dtts       

where the signal centroid, t0 , is

  )( 2 dttst t   0  . (9-5)  )( 2 dtts  If the signal is constant-envelope of duration T, (9-3) becomes 3  ( fˆ)  (9-6)  / NET 0int Equation (9-6) shows that all constant-envelope signals of a given duration and SNR have similar potential for accuracy of frequency measurement. Of course, realiza- tion of Cramér-Rao bound accuracy requires that the signal be known at the receiver and passed through a matched filter. This would be the case for a navigation-only transmission, but if frequency is to be measured from a communication data waveform, especially one involving considerable frequency or phase modulation, a decision-directed technique may be needed in which the demodulated data is used to reconstruct an estimated waveform that is subjected to frequency estimation. Thus any difference among the waveform candidates with respect to facilitating frequency measurement is primarily related to two things: signal strength and degree of waveform knowledge at the receiver. These differences are not expected to be substantial, and we can expect that all candidate modulation methods are somewhat equally amenable to frequency estimation. With regard to signal strength, the range of required b / NE 0 is between roughly 0 and 10 dB, from the most aggressive codes to uncoded BPSK or QPSK. This can be responsible for no more than a factor of ~ 3 difference in the frequency accuracy predicted by (9-7).

CMLP Study Page 96 August 2007 CMLP Final Report

9.2.1.3 Range and Doppler in CDMA

For a PN code of spread-spectrum chip rate Rc consisting of pseudo-randomly chosen binary chips, the ranging equation becomes 1 ( ˆ)  . (9-7) rIc NPTR 0 )/(

In (9-7), TI represents the integration, or observation time, and r / NP 0 is the received carrier-to-noise ratio. In this presentation we see that in theory any required signal- to-noise value can be achieved from a low r / NP 0 if the integration time can be made acceptably long. A limit on coherent integration time is placed by any uncer- ~ tainty in the received frequency. If the uncertainty is on the order of F Hz, coherent ~ integration over a span exceeding 4/(1 F) seconds becomes less productive because phase reversals between the incoming and reference signals tend to cancel the integration gain. If gain beyond that achievable by coherent integration is needed for reliable detection and/or estimation, noncoherent integration can be used to extend the effective integration interval. These are many means to do this, including the noncoherent accumulation of adjacent coherent integrations products, and data/Doppler-aided methods where estimates of data and frequency are used to construct a signal whose coherence can be maintained in the receiver for a longer period. To the extent that frequency uncertainty results from oscillator instability at either the transmit or receive location, the stability in the local frequency standards can become an issue in accurate registration of arriving signals. The corresponding formula for frequency measurement is 3  ( fˆ)  2/3 . (9-8)  rI / NPT 0 The chip rate, which determines the code bandwidth, is not a factor for frequency estimation as it is for arrival time estimation. Instead, duration is the driver, and the  2/3 dependence is as TI . Frequency error is related to speed (velocity magnitude) error,  (vˆ) , via the carrier frequency f0 as

( fˆ  () vˆ)  . f0 c (9-9)

CMLP Study Page 97 August 2007 CMLP Final Report

9.2.1.4 Navigation Embedded in or Based on Communication Sig- nals In CDMA communications, the raw material for communications and navigation are transmitted simultaneously. This approach is satisfactory as long as the processing gain of the CDMA system—ratio of chip rate to information symbol rate—exceeds a minimum, perhaps around 10 dB. The restriction to low data rates can be lifted by briefly interrupting, or puncturing, communication signals on occasion and with navigation-oriented signals, e.g., pseudo-noise (PN) codes whose chip rate is the symbol rate of the communications. This approach has interesting consequences. Radiometric navigation signals typically are transmitted at power levels such that the received r / NP 0 is small, with any SNR deficit being made up via integration time or code processing gain. When a bit received at b / NE o in the 0-10 dB range—depending on modulation and coding— becomes a chip of a PN code over which coherent integration can be performed, the resulting measurement SNR can be quite large. Revisiting the accuracy formulas under these assumptions leads to K ( ˆ)  (9-10) 3 bb NETR oI )/( and K  ( fˆ)  (9-11) 3 I bb NETR 0 )/( where K  1, and b / NE o represents the received bit SNR. These equations clearly illustrate how changes in data rate, observation time, carrier frequency and SNR impact the two raw measurement errors. For arrival time estimation, high data rate proves to be favorable for accuracy inasmuch as an increase in data rate is accompanied by both a power increase proportional to the rate, and a proportional bandwidth increase. For frequency estimation, integration time is the more important parameter. At a fixed data rate, increased integration time results in a greater RMS duration as well as greater total power. A similar analysis applies if the navigation measurements are based on the communication symbols themselves instead of an inserted code, with the proviso that the symbols must be demodulated reliably in order to form a good estimate of the transmitted waveform. Table 9-3 illustrates the type of capabilities required to meet some of the more demanding requirements of Table 9-1. The first entry shows that to meet a 10-m steady-state, 3 ranging error via a PN code, equation (9-7) demands a PN chip rate of about 16.2 MHz if the integrated SNR is 15 dB, a value typically associated with a

CMLP Study Page 98 August 2007 CMLP Final Report

good probability-of-detection/probability-of-false-alarm receiver operating characteristic (ROC). A second entry shows that if the SNR is increased another 14.7 dB by integrating about 30 times longer, the chip rate requirement becomes 3 MHz, the value currently used in the TDRSS SN. The additional integration time of course puts added pressure on the frequency accuracy required to support the measurement. For an absolute 3-D velocity accuracy of 2 cm/s, each component must be measured to accuracy 3 better than that, or 1.15 cm/s, 3. At S-band (we use the approximation f0 = 2 GHz) (9-8) says that to achieve the accuracy at 15 dB integrated SNR requires an integration time of about 381 ms and a carrier-to-noise ratio of dB 19.3 / NP or  19.3 dB . Raising the SNR to 20 dB yields requirements of 216 ms and 26.7 dB, respectively. The next examples in the table relate to the use of punctured or reconstructed communication signals as vehicles for extracting navigation information. This analysis leverages equations (9-10) and (9-11). We assume the communications achieves a bit SNR of b / NE 0 = 4 dB, representative of many coding techniques available today. To achieve 10 m accuracy at a data rate of 100 kbps, requires integration over 3.3 s, which is 330,000 bits. If the data rate is an order of magnitude greater, 1 Mbps, the duration drops to 3.3 ms, only 3300 bits. To measure speed to 2 cm/s by this means at 100 kbps requires 40 ms integration (40 bits). Increasing to 1 Mbps decreases the required to 18.3 ms. As the formula indicates, frequency accuracy is not nearly as sensitive to data rate as is range accuracy.

Capability (3) Method Key Parameters 10 m position PN code / NE 0int = 15.0 dB, Rc = 16.2 MHz

10 m position PN code / NE 0int = 29.7 dB, Rc = 3.0 MHz

2 cm/s speed PN code / NE 0int = 15.0 dB, TI = 381 ms, r / NP 0 = 19.3 dB

2 cm/s speed PN code / NE 0int = 29.0 dB, TI = 381 ms, r / NP 0 = 26.7 dB

10 m position Comm signal b / NE 0 = 4.0 dB, Rb = 100 kbps, TI = 3.3 s

10 m position Comm signal b / NE 0 = 4.0 dB, Rb = 1 Mbps, TI = 3.3 ms

2 cm/s speed Comm signal b / NE 0 = 4.0 dB, Rb = 100 kbps, TI = 40.0 ms

2 cm/s speed Comm signal b / NE 0 = 4.0 dB, Rb = 1 Mbps TI = 18.3 ms Table 9-3 Examples of Meeting Key Performance Requirements

CMLP Study Page 99 August 2007 CMLP Final Report

9.1

CMLP Study Page 100 August 2007 CMLP Final Report

10 Spectrum Constraints

As evidenced by the FOMs listed in Section 8, many factors are typically considered in the design and development of an efficient, high-performance satellite communications network. While many factors are considered, a subset typically emerges as the principal drivers of network design decisions. Invariably among these principal drivers are FOMs intended to satisfy spectrum-related requirements and constraints. This section provides a description of the spectrum-related requirements and constraints considered in the CMLP study and provides a rationale for their inclusion in this study.

The FOMs identified in Section 8 enable evaluation of CMLP techniques against a variety of spectrum-related requirements and constraints including:

 Relevant out-of-band (OOB) emission masks  International frequency allocation regulations  Spectrum efficiency guidelines  National and international Power Flux Density (PFD) regulations

Table 10-1 provides a top-level summary of the spectrum constraints and requirements considered in the CMLP study . It is important to emphasize that Table 10-1 is simply an overview and does not state all details of the spectrum requirements and constraints used in the CMLP study. The remainder of this section is intended to state the details of the spectrum requirements and constraints used in the CMLP study and provide a rationale or traceability for their use.

CMLP Study Page 101 August 2007 CMLP Final Report

Typical Maxi- Link Relevant OOB Spectrum Band mum Frequency PFD Levels(2) Direction Emission Masks Efficiency(1) Allocation Forward 6 MHz S NTIA, SFCG ≤ -154 dBW/m2 Return 6 MHz Forward 10 MHz ≥ 0.95 b/s/Hz(3) X NTIA, SFCG 150 MHz ≤ -150 dBW/m2 Return 10 MHz (DSN) Forward 25 MHz Ku NTIA ≤ -152 dBW/m2 Return 300 MHz Forward 50 MHz ≥ 0.95 b/s/Hz(4) Ka NTIA 1 GHz ≤ -115 dBW/m2 Return 650 MHz (SN) Notes: 1. Modulation spectrum efficiency, excludes effects of coding. 2. PFD limits stated for angles of arrival above the horizontal plane between 0 and 5. See Sec- tion 10.2.2.1 for PFD limits for other angles of arrival. 3. National and international spectrum rules, regulations and guidelines do not explicitly state a spectrum efficiency threshold or goal. The threshold stated here is based upon a precedent set forth by CCSDS modulation recommendations. 4. Although CCSDS recommendations are not applicable at Ku- and Ka-band, it is expected that the same degree of spectrum efficiency will be desired by the international consultative bodies at Ku and Ka-band as is currently recommended at S and X-band. Table 10-1: Spectrum Requirements and Constraints Considered in CMLP Study

10.1 Background

The spectrum requirements and constraints stated in Table 10-1 are traceable to or derived from the ITU-R Radio Regulations (rev. 2004), ITU-R Recommendations, Space Frequency Coordination Group (SFCG) Recommendations and Resolutions, and NTIA Regulations and Procedures. It should be noted that the constraints identified here are in many cases the direct result of other services sharing the bands used by Space Research Services and Space Operations Services. For example, in many bands Space Research Services share with Fixed Services. As a result there are power flux density (PFD) imposed on Space research services.

One caution is that constraints imposed by the ITU-R Radio Regulations, such as PFD limits, are more binding than constraints imposed by ITU-R recommendations or SFCG recommendations or resolutions. ITU-R recommendations do not have pe- nalties associated with non-compliance but are the basis for coordination and pro-

CMLP Study Page 102 August 2007 CMLP Final Report tection coordination between services, for example, filing for a frequency assignment with the ITU or NTIA or in an actual interference resolution scenario.

SFCG recommendations serve in a similar role between civilian space agencies. A waiver process for “extraordinary circumstances” is identified in SFCG but should not be relied upon to avoid SFCG constraints.

10.2 General Constraints

10.2.1 Spectrum Efficiency Spectrum efficiency is often quoted as a goal in successful spectrum management. A search of the Radio Regulations reveals that there is no definition of spectrum efficiency. There are however general definitions of efficiency (without the adjective spectrum applied) in the Radio Regulations and the Constitution.

Additionally there are many Resolutions and Recommendations that talk about efficiency. The definitions are arguably lacking in specifics and can lead to the conclusion that interpretation is subjective based upon practicality or the specific situation being addressed. 10.2.1.1 Specific Discussion of Efficiency as Applied to Bandwidth in the Radio Regulations The discussion contained in Article 3 of the Radio Regulations deals with bandwidth and is given below:

Article 3 Technical Characteristics of Stations 3.4 To the maximum extent possible, equipment to be used in a station should apply signal processing methods which enable the most efficient use of the frequency spectrum in accordance with the relevant ITU-R Recommendations. These methods include, inter alia, certain bandwidth expansion techniques, and in particular, in amplitude-modulation systems, the use of the single-sideband technique.

3.9 The bandwidths of emissions also shall be such as to ensure the most efficient utilization of the spectrum; in general this requires that bandwidths be kept at the lowest values which the state of the technique and the nature of the service permit. Appendix 1 is provided as a guide for the determination of the necessary bandwidth.

3.10 Where bandwidth-expansion techniques are used, the minimum spectral power density consistent with efficient spectrum utilization shall be employed.

3.11 Wherever necessary for efficient spectrum use, the receivers used by any service should comply as far as possible with the frequency tolerances of the

CMLP Study Page 103 August 2007 CMLP Final Report

transmitters of that service, due regard being paid to the Doppler effect where appropriate.

From the discussion provided in Article 3.9 above it could be suggested that narrow bandwidth leads to better spectral efficiency. However, Article 3.10 points out that bandwidth expansion techniques also lead to efficient spectrum utilization (again not defined) if the PFD is minimized. The conclusion here is that the Radio Regula- tions leave open the issue of how to achieve spectral efficiency and that even the term spectral efficiency is left undefined.

10.2.1.2 Conclusions on Spectrum Efficiency

Spectrum efficiency is not explicitly defined in the relevant ITU, NTIA, SFCG and CCSDS documentation. Correspondingly, the means of achieving spectral efficiency is also left undefined. In light of this lack of clear direction in the national and international spectrum regulations and guidelines, the CMLP team formulated a spectrum efficiency goal based upon the precedent set forth in the CCSDS modulation recommendations.

CCSDS recommends a variety of bandwidth efficient modulation techniques for Space Research Services operating at S-band and X-band. It can be shown that all CCSDS-recommended modulation techniques achieve a spectrum efficiency of 0.95 b/s/Hz or more. Considering the extensive analysis and deliberation that goes into CCSDS recommendations, the CMLP team felt very comfortable continuing a precedent set forth by CCSDS. Additionally, a required spectrum efficiency of ≥ 0.95 b/s/Hz preserves consideration of worthy candidate modulation techniques, such as filtered QPSK and OQPSK, while eliminating truly spectral inefficient modulation techniques such as BPSK.

10.2.2 Power Flux Density on Earth Constraints Imposed by the ITU-R Radio Regulations and ITU-R Recommendations Radio Regulations Article 21 imposes power flux density limits on terrestrial and space services sharing frequency band above 1 GHz. Of particular interest to the CMLP study team are the constraints these PFD limits place on Space Research Ser- vice (SRS), Space Operations (SO) and Intersatellite (ISS) including TDRSS and other services to protect Fixed & Mobile Services also operating in certain bands. As a positive consequence, these PFD limits also protect SRS Earth stations from Fixed Satel- lite Service SE links operating in these bands. These PFD limits were a major reason for TDRSS to be designed with spread spectrum in the 2025-2110 MHz and 2200- 2290 MHZ band. 10.2.2.1 PFD Limits from the Radio Regulations

CMLP Study Page 104 August 2007 CMLP Final Report

Table 10-2 provides the PFD limits as defined in Article 21 of the Radio Regulations. These PFD limits vary with angle of arrival on Earth and depend on a specific reference bandwidth given in Table 10-2.

Limit in dB(W/m2) for angles Reference Frequency of arrival (•) above horizontal band- Service band plane width 0º-5º 5º-25º 25º-90º Space Re- 2025-2110 search MHz Space Opera- -154 -154+0.5(•-5) -144 4 kHz 2200-2300 tions MHz (s-E) (s-s) 8025-8400 Space Re- -150 -150+0.5(•-5) -140 4 kHz MHz search (s-E) 22.55-23.55 ISS GHz Space Re- -115 -115+0.5(•-5) -105 1 MHz 25.25-27 search (s-E) GHz

Space Re- -120+0.75 37-38 GHz search (non- -120 -105 1 MHz • GEO) ( -5) Space Re- 37-38 GHz -125 -125+ (•-5) -105 1 MHz search (GEO)

Table 10-2: PFD Limits from Article 21 of the Radio Regulations

10.2.2.2 Example Showing the Need for Spread Spectrum to Reduce PFD on TDRSS Forward Link

In Table 10-3 the actual PFD is calculated and compared with the limits in 2025-2110 MHz forward link band for TDRSS using both un-spread BPSK and bandwidth spread CDMA. The example is meant to be illustrative of the need for spread spectrum to meet the PFD limits in the Radio Regulations and is not meant to indicate a real design for BER or user signaling design.

CMLP Study Page 105 August 2007 CMLP Final Report

In these tables, green means the value is a constant. Blue signifies an independent variable (value selected by the designer), while red signifies a dependent value (value calculated by a formula).

A simple comparison is made for a 100 Kb/s forward link from TDRSS to a LEO satellite assumed to be directly beneath the TDRSS satellite where the LEO has a 90 degree elevation angle and again when the LEO is near the limb of the Earth with a zero elevation angle to TDRSS.

Table 10-3 describes the signal structure used in the calculation. The 100 Kb/s data -12 signal requires an Eb/N0 of 5.5 dB to achieve a 10 BER

Low Rate Forward Service TDRSS to Earth Signal structure for service type information bit rate Mbps 0.10 dBHz 50.00 baseband filtering rolloff 0.20 occupied bandwidth MHz 0.27 dBHz 54.38 noise bandwidth MHz 0.23 dBHz 53.59 BER 1.00E-12 modulation BPSK bits per symbol 1.00 symbol rate MSps 0.23 coding RS/Viterbi inner code rate 0.50 outer code rate 223/255 net code rate 0.4373 Eb/No (theoretical) dB 4.00 implementation loss dB 1.50 Eb/No (required) dB 5.50 C/No (required) dBHz 55.50 C/N (required) dBHz 1.91 Table 10-3: GEO (TDRSS) to Earth S band Forward PFD

CMLP Study Page 106 August 2007 CMLP Final Report

Table 10-4 describes the TDRS to LEO forward link. The LEO satellite was assumed to have -2 dB antenna gain in the direction of TDRSS corresponding to a worst-case antenna gain for an omni antenna pattern with peaks and nulls. The EIRP of TDRSS -12 was adjusted to meet the specified performance of 5.5 dB Eb/N0 to achieve the 10 BER. This link shown here actually falls short of this requirement by about 0.03 dB.

CMLP Study Page 107 August 2007 CMLP Final Report

TDRS GEO to Earth frequency GHz 2.05 wavelength m 0.14624 antenna diameter m 4.80 antenna efficiency 0.55 antenna gain dB 37.67 antenna taper factor deg 70.00 antenna beamwidth deg 2.13 transmit power W 4.00 dBW 6.02 EIRP dBW 43.69

LEO satellite radius of Earth km 6378 orbital altitude km 400 antenna diameter Rx m 0.05 antenna efficiency 0.55 antenna gain dB -1.98 antenna taper factor deg 70 antenna beamwidth deg 204.74 satellite antenna noise temperature K 230.00 satellite receiver noise temperature K 100.00 system temperature K 330.00 dBK 25.19 G/T dB/K -27.16 Altitude of TDRSS km 35800.00 range to TDRSS km 35400 free space loss dB 189.66 dBW/K Boltzmann’s constant Hz -228.60 C/No (TDRSS to LEO satellite) dBHz 55.47 Eb/No TDRSS to LEO dB 5.47 Table 10-4: TDRS GEO to LEO Satellite Link Budget

Table 10-5 provides the resulting PFD on the Earth Using the TDRSS Link Budget in Table 10-4.

CMLP Study Page 108 August 2007 CMLP Final Report

TDRS EIRP dBW 43.69

Earth radius of Earth km 6378.00 TDRSS-to-Earth distance (minimum) at 90 deg el. Km 35400 TDRSS-to-Earth distance (maximum) at 90 deg el. Km 41630 spreading loss to Earth at 90 deg el dB 161.97 spreading loss to Earth at 0 deg el dB 163.38 Bandwidth for PFD KHz 4.00 PFD at 90 deg el on Earth in 4 KHz dBW/m^2 -132.26 PFD at 0 deg el on Earth in 4 KHz dBW/m^2 -133.67 PFD allowed by RR at 90 deg el on Earth in 4 KHz dBW/m^2 -144.00 PFD allowed by RR at 0 deg el on Earth in 4 KHz dBW/m^2 -154.00 actual PFD exceeds PFD of RR at 90 deg el without CDMA dBW/m^2 11.74 Amount PFD exceeds PFD limit at 0 deg el without CDMA dBW/m^2 20.33 PFD Reduction Using CDMA 3.02 Mc/s PN sequence in lieu of 100 Kb/s BPSK dB 14.80 Amount PFD exceeds PFD limit using a PN sequence at 90 deg elevation dB -3.06 Amount PFD exceeds PFD limit using a PN Se- quence at 0 deg elevation dB 5.53

Table 10-5: Resulting PFD

Without the use of spectral spreading the actual PFD exceeds the limit given by the Radio Regulations by some 11 to 20 dB. Using spectral spreading (CDMA) the limits are met at 90 degree elevation and for this example fall short by some 5.5 dB at 0 degree elevation. It should be noted that further changes would be necessary to meet the PFD for CDMA at 0 degree elevation for the example shown.

CMLP Study Page 109 August 2007 CMLP Final Report

10.2.2.3 Conclusions Related to PFD Limits in the Radio Regulations Because of limits on the PFD imposed by the Radio Regulations, care must be taken to ensure whatever modulation and coding selections are made by CMLP that these meet the PFD limits identified in Table 10-2. From the example given, it should be clear that without care these limits could be violated.

Although not analyzed here, emissions from LEO satellites communicating with TDRSS or with Earth stations may also violate the PFD limits because of their close proximity to Earth.

10.2.3 PFD Limit Relief for LEO and Data Relay Satellites The Recommendation entitled ITU-R SA. 1273 Power flux-density levels from the space research, space operation and Earth exploration-satellite services at the surface of the Earth required to protect the fixed service in the bands 2025-2110 MHz and 2200-2290 MHz may give a small amount of relief from the PFD limits contained in the Radio Regulations Article 21. This Recommendation still requires that a space station operating in 2200-2290 MHz the s-E direction meeting the PFD limits in Arti- cle 21, Radio Regulations. However the Recommendation also specifies that for a LEO satellite operating in 2200-2290 MHz in the s-s direction may be allowed a 3 dB higher PFD limit on Earth than allowed by the Radio Regulations.

A DRS meeting the Article 21 PFD limits in 2025-2110 MHz band will also meet the limits of the Recommendation, but the Recommendation identifies that the -130 dB(W/m^2) PFD limit corresponding to but not equivalent to the Radio Regulations limit, can be exceeded by up to 6 dB not more than 5% of the time to compensate for background interference. Note that the Recommendation uses 1 MHz as the reference bandwidth not 4 KHz which is given in the Radio Regulations which causes a seeming 24 dB increase in the recommended values in the Recommendation. See the Recommendation for further details. Also if there is a difference between the Rec- ommendation and the Article 21, Radio Regulations, the Radio Regulations will have priority.

10.2.3.1 Sharing in the 25.5-27 GHz Band

This Recommendation ITU-R 1625 imposes constraints on SRS satellites and their ability to create a PFD level on the GEO orbit where DRS satellites operate. While important, this constraint is not believed to drive the modulation and coding study.

10.2.3.2 PFD Limits in Deep Space Bands

CMLP Study Page 110 August 2007 CMLP Final Report

Recommendation ITU-R SA. 1157 Protection criteria for deep space research provides PFD limits for in band services and imposes constraints on other in band services sharing with deep space SRS.

Normally, these limits would not imply constraints on OOB emissions from adjacent bands , but by NASA convention and agreement these limits do impose constraints on near Earth missions in adjacent bands. For the purpose of the CMLP study these PFD limits may restrict the OOB emissions from near Earth missions in an adjacent band. It is recommended therefore that when any modulation or coding decision is made for any band adjacent to a Deep space band that the resulting OOB PFD be shown not to violate theses PFD limits for the Deep Space bands.

Recommended PFD limits for Deep Space Bands

1. that protection criteria for deep-space research Earth stations be established as follows » -222 dB(W/Hz) in bands near 2 GHz » -220 dB(W/Hz) in bands near 8 GHz » -220dB(W/Hz) in bands near 13 GHz » -216 dB(W/Hz) in bands near 32 GHz » 2. that protection criteria for stations on spacecraft in deep-space be established as follows; » -191 dB(W/20 Hz) in bands near 2 GHz » -189 dB(W/20 Hz) in bands near 7 GHz » -186 dB(W/20 Hz) in bands near 17 GHz » -184 dB(W/20 Hz) in bands near 34 GHz

3. that calculation of interference that may result from atmospheric and precipitation effects be based on weather statistics that apply for 0.001% of the time (see Para 2.3 Annex 1)

Constraints Imposed By the Space frequency Coordination Group (SFCG) Bandwidth Allocation Size and its Impact on Capacity

10.2.4 Other ITU-R Recommendations Certain other ITU-R Recommendations impose pfd limits on services other than SRS and Space Operations, which share in the bands. For example ITU-R SA. 609-2 Pro- tection criteria for radiocommunication links for manned and unmanned near-Earth research satellites. These other ITU-R Recommendations do not constrain SRS or SO Services except when a new modulation or coding approach is chosen by CMLP

CMLP Study Page 111 August 2007 CMLP Final Report

with increased susceptibility to interference. Such a choice could make the existing protection criteria insufficient to protect the selected CMLP modulation and coding.

In general selection of modulation and coding requiring increased sensitivity should be avoided since it is difficult to gain agreement on lower PFD limits in the ITU-R.

While these other ITU-R Recommendations are not believed to strongly influence the CMLP study the modulation and coding selected should be examined to ensure these ITU-R Recommendations do not require modification as a result of the modulation selection.

10.2.5 Constraints Imposed By the Space frequency Coordination Group (SFCG) 10.2.5.1 SFCG Resolutions

The SFCG is a coordination group composed of the world’s civil space agencies such as ESA, Russia, Japan and NASA. These agreements apply only to these agencies. The SFCG Handbook contains Resolutions and Recommendations that document theses agreements. A waiver from the constraints contained in these agreements is possible for “exceptional circumstances” with a sufficiently strong reason. The following Resolutions contain material thought to be relevant to the modulation and coding study.

10.2.5.1.1 Efficient use of spectrum in the 25.5-27 GHz and 37-38 GHz bands SFCG Res.19-1

This Resolution resolves that member agencies use bandwidth efficient modulation techniques whenever possible for high data rate space-to-Earth applications in the 25.5-27 GHz band and the 37-38 GHz band. The implication of this resolution seems to be that narrow bandwidth is a way to achieve spectral efficiency. The ambiguity between narrow bandwidth and bandwidth expansion leading to efficient design has already been addressed in Section 2.

10.2.5.1.2 Interference mitigation techniques for future systems planning to operate in the 2200-2290 MHz band SFCG Res. 24-1

This Resolution is an agreement for agencies to reduce interference in the S band by using common approaches to reducing interference.

This Resolution resolves that

CMLP Study Page 112 August 2007 CMLP Final Report

1. that systems developed for use in the 2200-2290 MHz band not transmit when beyond view of their cooperating Earth stations or when beyond view of their cooperating data relay systems;

2. that systems using this band be designed to minimize their bandwidths to reduce the potential interference to other systems in the band and that, whenever practical, bandwidths should not exceed 6 MHz, to reduce future congestion in the band;

3. that due consideration be given to interference mitigation techniques including Earth station geographical diversity, increased Earth station antenna gain enhancing the link margin, and reduced Earth station antenna sidelobe levels; 4. that other bands, like the band 25.5-27.0 GHz, be considered for high data rate systems.

10.2.5.2 SFCG Recommendations

10.2.5.2.1 Use of the 8450-8500 MHz band for space research SFCG Rec. 5- 1R5

This Recommendation provides direction for management of the near Earth (space- to Earth) portion of the X band and in particular imposes a 10 MHz limit on emissions in this band. Despite this band not being specifically selected by the SCAWG spectrum studies, the Recommendation is summarized below for completeness. The Recommendation states

1. that the 8450 - 8500 MHz band be used for Category A (Near Earth, or less than 2 million km from Earth) missions requiring an occupied bandwidth of up to 10 MHz per mission and having technical requirements that are best satisfied in the band;

2. that the band be used in particular for the mission to the Libration points with bandwidth requirements up to 10 MHz;

3. that utmost care be taken in the assignment of frequencies to these missions in order to make optimum use of the limited bandwidth available to Cat. A missions, and that the maximum bandwidth, postulated in “recommends 1” above, of 10 MHz per mission be strictly respected;

CMLP Study Page 113 August 2007 CMLP Final Report

4. that the 8450 - 8500 MHz not be used for Category B (Deep Space, or more than 2 million km from Earth) missions.

10.2.5.2.2 Protection of deep space research Earth stations from line of sight interference in the bands 2290-2300 MHz, 8400-8450 MHz and 31.8- 32,3 GHz SFCG Rec.14-1R1

This Recommendation establishes protection criteria (similar to ITU-R SA. 1157) which imposes constraints on other in band services sharing with deep space SRS and would normally not a constraint to SRS or the CMLP studies, but by NASA convention and agreement with SFCG this Recommendation imposes constraints on near Earth missions in adjacent bands.

Specifically this Recommendation states

1. that when a predicted interference potential exceeds the maximum interference power spectral flux density given in the following Table the provi- sions of SFCG RES A12-1 shall be applied;

2. that the values given in the Table apply for sources whether operating directly in-band or out of band and producing in-band spectral components.

CMLP Study Page 114 August 2007 CMLP Final Report

Frequency Maximum interference power spectral density flux density (dB(W/m2/Hz)) 2290-2300 MHz -257.0 8400-8450 MHz -255.1 31.8-32.3 GHz -249.3 Table of Allowable Interference Power Spectral Density

10.2.5.2.3 Efficient spectrum utilization for space science services on space- to-Earth links; Category A SFCG Rec. 21-2R2

This Recommendation deals with the bands 2290-2300 MHz, 8025-8400 MHZ EESS and the 8450-8500 MHz bands. The Recommendation proposes use of bandwidth efficient modulation, which is undefined and restricts the use of carriers and it imposes limits on the emission falling outside the main spectral component of the near Earth (space-to-Earth) bands.

The Recommendation specifically states

1. that, with immediate applicability to all space science service bands, space agencies use the most bandwidth efficient modulation schemes practicable for their missions;

2. that, with immediate applicability to all space science service bands, PCM/PM/Bi-phase or PCM/PM/NRZ modulation shall only be used when a carrier component is technically necessary and for symbol rates below 2 Ms/s.

3. that the emitted spectrum for all Space Science Services projects starting in/or after the year 2001 and that will utilize space-to-Earth link frequency assignments in the bands 2200–2290 MHz, 8025–8400 MHz and 8450–8500 MHz, adhere to the spectral emission masks in figure.

CMLP Study Page 115 August 2007 CMLP Final Report

Emission Relative to Peak (dB)

From SFCG 21-2R2 Efficient Spectrum Utilization for Space Science Services on Space-to Earth Links; Category A

10.2.5.2.4 Use of sub-carriers for space science services on space-to-Earth links; Cat.A SFCG Rec.21-3R1

This Recommendation restricts the use of sub carriers that typically increase the occupied bandwidth in near Earth mission bands.

CMLP Study Page 116 August 2007 CMLP Final Report

1. that, with immediate applicability to all space science service bands Cat.A, sub-carrier modulation shall not be used except where absolutely required and then only for symbol rates below or equal to 60 kb/s;

2. that, with immediate applicability to all space science service bands Cat. A, if a sub-carrier is required, it shall comply with the specifications set forth in considering e) and f);

10.2.5.2.5 Efficient spectrum utilization for space research service, deep space (Category B), in the space-to-Earth link SFCG Rec. 23-1

This Recommendation deals with the band Deep Space downlink band 8400-8450 MHz and restricts the maximum allowable bandwidth to not exceed 8 MHz for Mars missions and 12 MHz for non-Mars missions. This Recommendation also imposes a maximum PFD limit at Earth in this band. Specifically the Recommendation states

1. that, in the 8400-8450 MHz band, the maximum allowable bandwidth of telemetry signals be limited according to Figure 1, wherein a) the lower curve applies to all missions; b) the upper curve applies only to the non-Mars-missions, strictly on condition that they would not interfere with the Mars missions;

2. that, in the 8400-8450 MHz band, the spectral power flux density outside the maximum allowable bandwidth be limited to –266 dB(W/Hz/m^2) on the surface of the Earth;

3. that member agencies use the 32 GHz-band for high rate telemetry with bandwidth requirement higher than those allowed in Figure 1; (See next chart.)

4. that except for scientific or technical reasons, subcarrier frequencies above 60 kHz do not exceed 5 times the maximum symbol rate of the mission and do not exceed 300 kHz.

CMLP Study Page 117 August 2007 CMLP Final Report

10.2.5.2.6 Use of 37-38 GHz space research service allocation SFCG Rec. 14- 2R5 This Recommendation establishes guidelines for the use of this band for manned missions for deep space and Lunar missions.

The lower 500 MHz of the band has priority for deep space but can be used for lunar missions if this use is not incompatible with manned planetary missions. The upper 500 MHz is the primary choice for manned Lunar missions. No PFD limits are imposed.

Specifically this Recommendation states

1. that the 37-37.5 GHz band be maintained available for implementation of space-to-Earth links for manned and unmanned planetary missions and for development and operation of manned planetary missions in the Lu- nar environment, recognizing that manned missions have priority than unmanned missions;

2. that Earth-to-space links for manned lunar and manned and unmanned planetary exploration be implemented in the band 40-40.5 GHz or other Earth-to-space bands as appropriate;

CMLP Study Page 118 August 2007 CMLP Final Report

3. that to protect the manned planetary missions, all incompatible lunar missions cease their operations when manned planetary missions are present in the deep space environment;

4. that sun-Earth libration point (L2) missions considering to use the 37-38 GHz band implement their space-to-Earth links in the 37.5-38 GHz portion of the band, with associated Earth-to-space links in the 40-40.5 GHz band or other Earth to space bands as appropriate;

5. that Space VLBI systems implementing time-critical downlinks requiring up to 1 GHz of real-time bandwidth utilize the band 37-38 GHz, recognizing the need for operational coordination, when required, with manned lunar and planetary exploration systems

6. that Cat. A Space Research service missions, that can share with FSS, be accommodated in the 37.5-38 GHz portion of the band with associated Earth-to-space links in appropriate bands;

7. that Member agencies take into account the information contained in the Annex when examining intra-service sharing in the 37-38 GHz band.

10.2.5.2.7 Assignment of differential one-way ranging tone frequencies for Category B missions SFCG Rec. 23-2

This Recommendation establishes a methodology for increasing available Deep Space bandwidth by utilizing adjacent unallocated bands for ranging signals by im- posing PFD limits in bands adjacent to the allocated deep space 8400-8450 MHz and 32.8-32.3 GHz bands so as to not cause interference to other services in the Earth vicinity. The approach excludes deep space emissions in the adjacent Radio Astronomy band of 31.3-31.8 GHz.

Specifically the Recommendation states

1. that member agencies assign DOR tone frequencies within the existing Cat. B allocations whenever possible;

2. that member agencies, when it is necessary to assign a DOR tone frequency outside a Cat. B allocation.limit the Power Flux Density of each tone to -211 dB(W/m^2) in the 8 Ghz band and -204 dB(W/m^2) in the 32 GHz Band;

CMLP Study Page 119 August 2007 CMLP Final Report

3. that member agencies do not assign DOR tones (including intermodulation products when multiple tone pairs are used simultaneously) in the 31.3-31.8 GHz band.

10.2.6 Other Relevant Emission Masks Table 10-6 provides the NTIA OOB emission mask.

Table 10-6: NTIA OOB Emission Mask

10.3 Conclusions

The spectrum material provided here imposes constraints on the modulation and coding study (CMLP). In particular, the PFD limits in the Radio Regulations should be met. Special care should be taken to ensure that the PFD limits described in Sec- tion 10.2.3 and which are easily violated for Data Relay Satellites and near Earth missions are met. This may require in some cases using spectral spreading.

CMLP Study Page 120 August 2007 CMLP Final Report

The signaling bandwidth limitations identified in the 2 GHz and 8 GHz bands should be adhered to.

The CMLP should avoid choosing modulation and coding schemes having increased sensitivity to interference requiring a decrease in the allowable interference PFD limits in the various ITU-R and SFCG Recommendations.

CMLP Study Page 121 August 2007 CMLP Final Report

11 CMLP Selection

11.1 Process

Table 11-1 provides a top-level overview of the process used to derive coding, modulation, multiple access and link protocol recommendations for future NASA communications systems. A FOM-based down-select process is used which relies upon a variety of critical inputs and utilizes supporting technical analysis to evaluate the candidate CMLP techniques.

Input Process Output

CMLPCMLP Catalogs Catalogs CMLPCMLP Team Team Review Review RecommendedRecommended •Links TechniquesTechniques • Modulations • Codes NASA Review • MA Techniques NASA Review • By link class • Link Protocols

CMLPCMLP FOM FOM Define Identify Rank FOMs Perform Initial Perform List Define Identify Rank FOMs Perform Initial Perform List EvaluationEvaluation DrivingDriving DownselectDownselect FinalFinal ScenariosScenarios RequirementsRequirements DownselectDownselect Report ExistingExisting Report • Define space •For the Information • On a link-by- Information comm • Rank FOMs remaining Sources link basis and Sources evaluation for each techniques, against a scenarios • Identify driving evaluation perform a • CCSDS handful of requirements scenario and detailed • Textbooks critical FOMs, • Formulate top- based upon link type evaluation • Papers perform initial level ops best under- against all downselection concept for standing of • Different link applicable each eval future mission classes will FOMs •Critical FOMs scenario needs have different ExistingExisting typically needs NASA include power NASA • Identify links Techniques efficiency, Techniques associated spectral with each eval SupportingSupporting efficiency, Agency scenario Analysis Agency robustness, etc Analysis MissionMission PlanningPlanning ModelModel

Regulatory Regulatory Generate Documentation IssuesIssues Generate Documentation

Table 11-1: Top-Level CMLP Down-select Process

11.1.1 Inputs

The integrity of the down-select process is dependent upon the accuracy of the inputs. The CMLP group has worked diligently to ensure the inputs are current, complete and accurate. It is acknowledged that some of the inputs used for this study may change over time, for example, the mission model and link catalog, however, it is not expected that the recommendations of this report will change given reasonable evolution in the various inputs.

CMLP Study Page 122 August 2007 CMLP Final Report

Section 5 of this report provides the CMLP link catalog which served as an input to the CMLP down-selection process. The catalog provides a representative set of communication links expected to be required by future NASA missions. The CMLP link catalog does not explicitly include communication links associated with future international missions, however, the stated set is expected to represent potential future international missions.

The link catalog has the potential to change and evolve more than probably any other input to the CMLP down-select process. Fortunately, the link catalog is so extensive that modest changes in the various links likely should not alter the general composition and classes of links considered by this study.

Section 7.1 provides the CMLP modulation catalog. All known candidate space communication modulations are identified and the associated characteristics and performance described. The CMLP down-select process considered all of these modulations as viable candidates for the various evaluation scenarios.

Section 7.2 provides the CMLP coding catalog. As with the modulation catalog, the coding catalog contains all known candidate space communication codes and describes the associated characteristics and performance.

Section 7.3 provides the CMLP Multiple Access (MA) catalog. Section 7.4 provides the CMLP Link Protocol (LP) catalog.

Section 8 of this report provides the FOMs that served as an input to the CMLP down-selection process. The FOMs were identified and defined by the entire CMLP team. Due to the importance of these FOMs in the down-select process great effort was placed on arriving at FOMs that would enable a comprehensive and appropriate evaluation of all of the CMLP techniques considered in this study.

Although not duplicated here, the entire library of CCSDS documents and associated recommendations served as an input to the down-select process. While no requirement was levied on the CMLP process to adhere to CCSDS recommendations, the CMLP team recognized the importance and appropriateness of the recommendations and awarded credit to techniques that were CCSDS compliance through the FOM titled Alignment with International Standards.

Also serving as an input to the down-select process are the CMLP techniques currently supported by NASA communications networks. Sound judgment went into the specification, design and implementation of current NASA and international agency networks. The techniques used by current networks have an extensive legacy of successful and efficient use that cannot be discarded. While no requirement was levied on the CMLP down-select process to adhere to the techniques currently in use

CMLP Study Page 123 August 2007 CMLP Final Report by NASA networks, credit was awarded to legacy techniques through the FOMs titled Supports Legacy Missions and Infrastructure Burden.

To ensure that the recommendations emerging from the down-select process were compliant with all national and international regulations, regulatory requirements such as the NTIA emission mask and ITU spectrum requirements served as inputs to the CMLP down-select process.

11.1.2 Down-select Process

The down-select process begins with the definition of the top-level communication network evaluation scenarios. To accomplish this step, future required communication scenarios must be envisioned and an operations concept drafted. If sufficient similarity exists, scenarios can be grouped so as to limit the amount of evaluation scenarios.

The CMLP team identified five top-level communication system evaluation scenarios: Near Earth Relay, Near Earth Direct To Earth / Direct From Earth (DTE/DFE), Lunar Relay, Mars Relay and Deep Space DTE/DFE. Section 5 provides additional insight into the communication network evaluation scenarios identified as part of the CMLP study.

As part of defining the evaluation scenarios, it is necessary to determine the simultaneous links, link duty cycles, ranging requirements (if required) and quality of service requirements. It is also required that candidate implementation scenarios be devised, such as, near Earth DTE/DFE ground terminal antenna size to understand the discrimination levels present in an evaluation scenario.

The next step in the down-select process is to identify driving requirements. These are requirements that have the potential to dramatically impact the selection of appropriate CMLP techniques. For this CMLP study, driving requirements were identified to include the need to support ranging, quality of service, simultaneous links, latency, among others.

Following definition of the evaluation scenarios and identification of the driving requirements, the FOMs must be ranked in terms of importance for each down-select area, where down-select areas include modulation, coding, MA and link protocol. Ranking of these requirements are unique to each down-select area. What may be important to the MA down-select may not be important to the coding down-select. For instance, support of ranging is important for the MA down-select, however, is largely not applicable to the coding down-select.

CMLP Study Page 124 August 2007 CMLP Final Report

Sections 11.3, 11.4, 11.5 and 11.6 provide the CMLP-derived FOM rankings for the modulation, coding, MA and link protocol down-selects, respectively. The rankings shown in the tables can be debated, however, generally consensus agreement existed within the CMLP group regarding the rankings shown in the tables.

Having weighted the FOMs appropriately for each down-select area, the FOMs that are clearly the most important to each down-select area are used to perform the initial down-select for that down-select area. This initial down-select eliminates techniques that are clear underperformers and eliminates them from further consideration. The intent of this step is to reduce the field of candidate techniques, thereby, enabling the maximum amount of analysis to be focused on the techniques that truly offer the best and most appropriate performance.

Having thinned the field of candidate CMLP techniques, an exhaustive down-select process is applied to these remaining techniques against all of the applicable CMLP FOMs. The performance of each candidate technique is assessed relative to the other competing candidate techniques. Performance scores are computed for each candidate technique considering the performance of these techniques in each of the applicable FOM areas and considering the FOM weightings.

11.1.3 Outputs

The output of the CMLP down-select process is a set of CMLP recommendations for each link class. Recommendations formulated through this process are documented in Section 12 of this report.

11.2 Code and Modulation Interdependence

We have developed a methodology [Appendix X] for estimating and comparing the performance of coded modulation systems for a very wide range of codes and modulations. Performance thresholds (in decibels) for a given code and modulation are decomposed into four major components:

1) the BI-AWGN capacity threshold, determined for the binary-input additive white Gaussian noise channel at its capacity limit;

2) a correction called the modulation penalty with respect to BPSK, giving the difference between the capacity thresholds for the given modulation and for BPSK;

3) a correction called the finite-size penalty that accounts for the inability of any finite-size coded system to approach the capacity limits fully;

CMLP Study Page 125 August 2007 CMLP Final Report

4) a residual term called the size-constrained non-optimality of the given code and modulation.

The usefulness of this decomposition is that the fourth term, the residual size- constrained non-optimality, can generally be made uniformly small (0.5 dB to 1.0 dB) with a well-designed coded modulation system. Larger and more significant variations in the performance thresholds are predictable from the first three terms, which are computable without resort to simulations and which depend on the major parameters of the code and modulation such as code size, code rate, and modulation rate and type. This in turn enables a simpler and more insightful analysis and evaluation of fundamental tradeoffs between important system parameters such as power efficiency, bandwidth efficiency, and latency, that are impacted by the choice of code and modulation.

For the purposes of the CMLP study, coded modulation schemes that rank poorly based on the fourth term (their size-constrained non-optimalities) were deemed to be inherently inferior in power efficiency to those with size-constrained non- optimalities in the 0.5 dB to 1.0 dB range. Such schemes would not be recommended except on the basis of other FOMs (such as complexity).

The third term (the finite-size penalty) is to a good approximation nearly independent of the modulation and the rate of the code. This term summarizes the fundamental tradeoffs between code size and power efficiency, at different desired error rate levels. In the CMLP down-select process, acceptable code sizes and error rates are determined by application-specific latency and data fidelity requirements. Thus, for CMLP purposes the finite-size penalty component is purely informative, in the sense that it estimates the price in power efficiency paid to satisfy the latency and fidelity constraints but it does not distinguish between two coded modulation systems satis- fying the same constraints.

The first two terms together estimate the power efficiency for a coded modulation system using a given modulation and a code of a given rate, assuming no latency constraints (infinitely long code) and arbitrarily low error rate. In the CMLP down- select process, these two terms are most useful in evaluating the fundamental tradeoffs between power efficiency and bandwidth efficiency for the various bandwidth constraints imposed by the representative communication links considered for the CMLP study. The two terms separately are useful in distinguishing the parts of this tradeoff that are primarily due to the choice of code rate and those that are primarily due to the choice of modulation.

CMLP Study Page 126 August 2007 CMLP Final Report

11.3 Modulations

Table 11-2 provides the FOM rankings as formulated by the CMLP team for the Cat- egory A mission modulation down-select. As can be seen from the table, power efficiency, spectrum efficiency, user burden, robustness and hardware maturity are the FOMs assigned the greatest weighting.

Table 11-3 provides the FOM rankings as formulated by the CMLP team for the category B mission modulation down-select. Similar trends exist in the Category B mission FOM rankings as in the Category A FOM rankings, however, for Category B missions less emphasis needs to be placed on spectral efficiency.

Considering just the FOMs that were assigned the greatest weighting, an initial modulation down-select was performed for the Category A GN and SN links, Cate- gory A Constellation links and Category B links. The objective of this initial down- select was to eliminate modulation techniques that dramatically underperformed in key FOM areas.

Table 11-4 provides insight into the initial down-select process and results for Cate- gory A near-Earth SN and GN links. Table 10-5 and Table 11-6 provide insight into the initial down-select process and results for Category A Constellation links and Category B links, respectively.

Having reduced the trade-space to a manageable level through the initial down- select, a comprehensive evaluation was performed against all FOMs for the modulation techniques that advanced through the initial down-select. The comprehensive evaluation is based upon existing analysis or analysis performed in support of the CMLP study. In some FOM areas, however, a subjective evaluation had to be performed.

Table 11-7, Table 11-8 and Table 11-9 provide the final down-select evaluation results for Category A near-Earth SN and GN links, Category A Constellation links and Category B links, respectively.

Because of the similarity between Category A Constellation link scenarios and Mars Relay link scenarios, the modulation evaluation and conclusions given in Table 11-5 and Table 11-8 are proposed for applicability to the Mars Relay scenario.

CMLP Study Page 127 August 2007 CMLP Final Report

Ranking Figure of Merit 1 = Most important 5 = Least important N/A = Not applicable

Supports legacy missions (time span, percent of features) -This is more than a statement about supporting missions that are currently flying. At any point in time, this FOM is still important. For example, in 5 2025 we will still need to support missions launched in 2020. Hence, we view this FOM as a sliding window of support. The size of the window defines the metric, along with the number of legacy features that are supported.

Spectrum utilization -This FOM reflects spectrum constraints, such as channel bandwidth (Hz) and Power Flux Density (PFD) limits (W/m2), and bandwidth efficiency 2 (bits/sec/Hz). It includes the use of algorithms to prioritize data to get the best use out of the available spectrum.

Power efficiency (PT/(RN0) to get required performance) -Here PT is the total power (the sum of modulated data power Pd and residual carrier power Pc, if any), R is the data rate, and N0 is the noise spectral density. Note that PT/(RN0) = (Pd+Pc)/(RN0) = Eb/N0 + Pc/(RN0). Thus, for suppressed carrier modulations (i.e., Pc=0), PT/(RN0) is 1 identical to the familiar power efficiency measure Eb/N0, while for residual carrier modulations (i.e., Pc>0), PT/(RN0) correctly accounts for the additional power used in the carrier. We will also allow for prioritization schemes as discussed in “spectrum efficiency.)

User burden (Percent cost increase) -This FOM reflects the additional costs a mission user will incur to use the scheme in question. It includes costs due to mass, power, and 2 spacecraft components. It also includes costs for real time operation and for mission planning functions.

Infrastructure burden (Percent cost increase) -This is similar to “user burden” but the Space Communications and Navigation (SCAN) infrastructure elements incur these costs. It is critical to 3 measure the required cost increases on both sides in order to make programmatic decisions on investment.

Alignment with international standards (Probability of alignment) -Alignment is very difficult to measure. We decided to measure this as a subjective probability that the recommended scheme will become an 4 international standard. If, for example, all aspects are already standards, then this FOM is a 1. If there are elements that we believe can never become standards, it is a 0. International standards includes those managed by CCSDS, ITU/SFCG, IOAG, and IEEE.

Provide radiometrics for navigation (Accommodation % cost increase) -Radiometrics may be required on various links to provide navigation or radio science. We will assume appropriate requirements for these N/A dependant on the link. The data types being considered will include, Doppler, Doppler rate, ranging, and differential techniques, as appropriate. We will measure the additional cost (as a percent) to accommodate the required data types at the required performance levels.

Robustness -This FOM considers robustness to short term signal disruptions, whether caused by mechanical failure (e.g. inadvertently spinning spacecraft) or by sensitivity to electronic signal distortions. This FOM also considers the effects of interference, including electromagnetic interference (EMI), 2 inter-symbol Interference (ISI), and cross-channel interference (CSI). We will measure the additional power required to mitigate these sources to the levels required by the individual link classes.

Latency (seconds) -Latency will come from at least three major sources: end-to-end latency, time to acquire signals, and latency from signal processing. These are 5 somewhat independent. The sum of these will be used for this FOM.

Technology maturity (TRL) 2 -We will estimate the technology readiness level (TRL) of the critical components of the system and make a judgment as to the overall maturity.

Capacity -We will consider the aggregate capacity of simultaneous links between multiple elements, including the data rates that can be supported N/A between individual elements and the number of simultaneous links. Table 11-2: FOM Rankings for Category A Mission Modulation Down-select

CMLP Study Page 128 August 2007 CMLP Final Report

Ranking Figure of Merit 1 = Most important 5 = Least important N/A = Not applicable

Supports legacy missions (time span, percent of features) -This is more than a statement about supporting missions that are currently flying. At any point in time, this FOM is still important. For example, in 4 2025 we will still need to support missions launched in 2020. Hence, we view this FOM as a sliding window of support. The size of the window defines the metric, along with the number of legacy features that are supported.

Spectrum utilization -This FOM reflects spectrum constraints, such as channel bandwidth (Hz) and Power Flux Density (PFD) limits (W/m2), and bandwidth efficiency 3 (bits/sec/Hz). It includes the use of algorithms to prioritize data to get the best use out of the available spectrum.

Technology maturity (TRL) 2 -We will estimate the technology readiness level (TRL) of the critical components of the system and make a judgment as to the overall maturity.

Capacity -We will consider the aggregate capacity of simultaneous links between multiple elements, including the data rates that can be supported N/A between individual elements and the number of simultaneous links. Table 11-3: FOM Rankings for Category B Mission Modulation Down-select

CMLP Study Page 129 August 2007 CMLP Final Report

Remaining Modulations after Link Description (1,2) First Stage Downselect First Stage Downselect Process Link Name (Eliminate modulations which underperform

Modulation Shaping/Filter on certain critical FOMs) Link Type Data Rate ID Type

Precoded GMSK (h = 0.5) Gaussian (BT_b = 0.5) - Req'd Eb/No at 1E-5 BER <= 11.0 dB - Req'd bandwidth (considering all potential applicable SRRC (roll-off factor = 1.0) codes) <= 6 MHz (standard S-band allocation) - Spectral efficiency >= 0.95 using 99% bw (i.e., meet the OQPSK (SQPSK) NTIA out-of-band emission mask) SRRC (roll-off factor = 0.5) - High hardware maturity required (i.e., keep forward operational link simple and low risk; no trellis receiver Operational possible) ≤ 60 kbps Butterworth 6th order Forward - Special exceptions to this process have been made for OQPSK/PM modulations which help ensure carr acq at low C/No, i.e., Bessel 6th order PCM/PSK/PM and PCM/PM/NRZ - Although not recommended here, differential PSK modulations should be considered in cases where a low- PCM/PSK/PM TBD complexity, high-reliability link design is required - this safety is at the expense of worse BER performance relative to modulations recommended by the downselection process PCM/PM/NRZ TBD

Gaussian (BT_b = 0.5) Precoded GMSK (h = 0.5) Gaussian (BT_b = 0.25) - Req'd Eb/No at 1E-5 BER <= 11.0 dB GMSK (h = 0.5) Gaussian (BT_b = 0.3) - Req'd bandwidth (considering all potential applicable codes) <= 6 MHz (standard S-band allocation) SRRC (roll-off factor = 1.0) - Spectral efficiency >= 0.95 using 99% bw and assuming a OQPSK (SQPSK) SRRC (roll-off factor = 0.5) rate 1/2 code (i.e., meet the NTIA out-of-band emission mask) SRRC (roll-off factor = 0.2) S-band - Medium or high hardware maturity required (i.e., keep Operational return operational link relatively simple and low risk) ≤ 60 kbps Butterworth 6th order Return OQPSK/PM - Special exceptions to this process have been made for Bessel 6th order modulations which help ensure carr acq at low C/No, i.e., PCM/PSK/PM and PCM/PM/NRZ FQPSK-B Defined by modulation - Although not recommended here, differential PSK modulations should be considered in cases where a low- Defined by modulation SOQPSK-A complexity, high-reliability link design is required - this safety is at the expense of worse BER performance relative to SOQPSK-B Defined by modulation modulations recommended by the downselection process PCM/PSK/PM TBD

PCM/PM/NRZ TBD Gaussian (BT_b = 0.5) Precoded GMSK (h = 0.5) Gaussian (BT_b = 0.25) GMSK (h = 0.5) Gaussian (BT_b = 0.3) SRRC (roll-off factor = 1.0) - Req'd Eb/No at 1E-5 BER <= 11.0 dB - Req'd bandwidth (considering all potential applicable OQPSK (SQPSK) SRRC (roll-off factor = 0.5) codes) <= 6 MHz (standard S-band allocation) Operational / Science > 60 kbps SRRC (roll-off factor = 0.2) - Spectral efficiency >= 0.95 using 99% bw and assuming a Return rate 1/2 code (i.e., meet the NTIA out-of-band emission Butterworth 6th order OQPSK/PM mask) Bessel 6th order - Medium or high hardware maturity required

FQPSK-B Defined by modulation SOQPSK-A Defined by modulation SOQPSK-B Defined by modulation Table 11-4: Category A SN and GN Link Modulation Initial Down-select Results

CMLP Study Page 130 August 2007 CMLP Final Report

Remaining Modulations after Link Description (1,2) First Stage Downselect First Stage Downselect Process Link Name (Eliminate modulations which underperform

Modulation Shaping/Filter on certain critical FOMs) Link Type Data Rate ID Type

Precoded GMSK (h = 0.5) Gaussian (BT_b = 0.5) - Req'd Eb/No at 1E-5 BER <= 11.0 dB - Req'd bandwidth (considering all potential applicable SRRC (roll-off factor = 1.0) codes) <= 10 MHz (standard X-band allocation) - Spectral efficiency >= 0.95 using 99% bw (i.e., meet the OQPSK (SQPSK) NTIA out-of-band emission mask) SRRC (roll-off factor = 0.5) - High hardware maturity required (i.e., keep forward operational link simple and low risk; no trellis receiver possible) ≤ 60 kbps Butterworth 6th order - Special exceptions to this process have been made for OQPSK/PM modulations which help ensure carr acq at low C/No, i.e., Bessel 6th order PCM/PSK/PM and PCM/PM/NRZ - Although not recommended here, differential PSK Operational modulations should be considered in cases where a low- Forward PCM/PSK/PM TBD complexity, high-reliability link design is required - this safety is at the expense of worse BER performance relative to modulations recommended by the downselection process PCM/PM/NRZ TBD

Precoded GMSK (h = 0.5) Gaussian (BT_b = 0.5) - Req'd Eb/No at 1E-5 BER <= 11.0 dB SRRC (roll-off factor = 1.0) - Req'd bandwidth (considering all potential applicable OQPSK (SQPSK) codes) <= 10 MHz (standard X-band allocation) > 60 kbps SRRC (roll-off factor = 0.5) - Spectral efficiency >= 0.95 using 99% bw (i.e., meet the NTIA out-of-band emission mask) Butterworth 6th order - High hardware maturity required (i.e., keep forward OQPSK/PM operational link simple; no trellis receiver possible) X-band Bessel 6th order

Gaussian (BT_b = 0.5) Precoded GMSK (h = 0.5) Gaussian (BT_b = 0.25) GMSK (h = 0.5) Gaussian (BT_b = 0.3) SRRC (roll-off factor = 1.0) - Req'd Eb/No at 1E-5 BER <= 11.0 dB - Req'd bandwidth (considering all potential applicable OQPSK (SQPSK) SRRC (roll-off factor = 0.5) codes) <= 150 MHz (standard X-band allocation) Operational / Science > 60 Mbps and ≤ 125 Mbps SRRC (roll-off factor = 0.2) - Spectral efficiency >= 0.95 using 99% bw (i.e., meet the Return NTIA out-of-band emission mask) Butterworth 6th order OQPSK/PM - Medium or high hardware maturity required (i.e., keep Bessel 6th order return operational link relatively simple and low risk)

FQPSK-B Defined by modulation SOQPSK-A Defined by modulation SOQPSK-B Defined by modulation

SRRC (roll-off factor = 0.5) - Req'd bandwidth (considering all potential applicable 8PSK codes) <= 150 MHz (maximum X-band allocation); this Science basically mandates a spectral efficiency >= 2.3 using 99% 300 Mbps SRRC (roll-off factor = 0.35) Return bw and assuming a rate 7/8 code - Req'd Eb/No at 1E-5 BER <= 14.0 dB - Peak-to-Average Power Ratio < 5 dB 16-APSK (12,4) SRRC (roll-off factor = 0.5)

Table 11-4: Category A SN and GN Link Modulation Initial Down-select Results (Continued 1)

CMLP Study Page 131 August 2007 CMLP Final Report

Remaining Modulations after Link Description (1,2) First Stage Downselect First Stage Downselect Process Link Name (Eliminate modulations which underperform on certain critical FOMs) Modulation Shaping/Filter Link Type Data Rate ID Type

Gaussian (BT_b = 0.5) Precoded GMSK (h = 0.5) Gaussian (BT_b = 0.25)

GMSK (h = 0.5) Gaussian (BT_b = 0.3)

SRRC (roll-off factor = 1.0) - Req'd bandwidth (considering all potential applicable OQPSK (SQPSK) SRRC (roll-off factor = 0.5) codes) <= 300 MHz (maximum Ka-band allocation); this basically mandates a spectral efficiency >= 0.95 using 99% Ku-band Return ≤150 Mbps SRRC (roll-off factor = 0.2) bw and assuming a rate 7/8 code - Req'd Eb/No at 1E-5 BER <= 11.0 dB Butterworth 6th order OQPSK/PM - Medium or high hardware maturity required Bessel 6th order

Defined by modulation FQPSK-B Defined by modulation SOQPSK-A Defined by modulation SOQPSK-B Precoded GMSK (h = 0.5) Gaussian (BT_b = 0.25) GMSK (h = 0.5) Gaussian (BT_b = 0.3) SRRC (roll-off factor = 1.0) OQPSK (SQPSK) SRRC (roll-off factor = 0.5) - Req'd bandwidth (considering all potential applicable codes) <= 650 MHz (maximum Ka-band allocation); this Science SRRC (roll-off factor = 0.2) basically mandates a spectral efficiency >= 1.1 using 99% ≤ 600 Mbps Return Butterworth 6th order bw and assuming a rate 7/8 code OQPSK/PM - Req'd Eb/No at 1E-5 BER <= 11.0 dB Bessel 6th order - Medium or high hardware maturity required FQPSK-B Defined by modulation Ka-band SOQPSK-A Defined by modulation SOQPSK-B Defined by modulation

OQPSK (SQPSK) SRRC (roll-off factor = 0.2) - Req'd bandwidth (considering all potential applicable codes) <= 650 MHz (maximum Ka-band allocation); this SRRC (roll-off factor = 1.0) Science basically mandates a spectral efficiency >= 1.75 using 99% 1 Gbps Return bw and assuming a rate 7/8 code 8PSK SRRC (roll-off factor = 0.5) - Req'd Eb/No at 1E-5 BER <= 13.3 dB - Medium or high hardware maturity required SRRC (roll-off factor = 0.35) Table 11-4: Category A SN and GN Link Modulation Initial Down-select Results (Continued 2)

CMLP Study Page 132 August 2007 CMLP Final Report

Remaining Modulations after Link Description (1,2) First Stage Downselect First Stage Downselect Process Link Name (Eliminate modulations which Data Rate Modulation Shaping/Filter underperform on certain critical FOMs) Link Type (kbps) ID Type

Precoded GMSK (h = 0.5) Gaussian (BT_b = 0.5) - Req'd Eb/No at 1E-5 BER <= 11.0 dB - Req'd bandwidth (considering all potential applicable SRRC (roll-off factor = 1.0) codes) <= 6 MHz (standard S-band allocation) - Spectral efficiency >= 0.95 using 99% bw (i.e., meet OQPSK (SQPSK) the NTIA out-of-band emission mask) SRRC (roll-off factor = 0.5) - High hardware maturity required (i.e., keep forward link simple and low risk; no trellis receiver possible) - Special exceptions to this process have been made for Forward/Uplink ≤ 60 Butterworth 6th order modulations which help ensure carr acq at low C/No, OQPSK/PM i.e., PCM/PSK/PM and PCM/PM/NRZ Bessel 6th order - Although not recommended here, differential PSK modulations should be considered in cases where a low- complexity, high-reliability link design is required - this PCM/PSK/PM TBD safety is at the expense of worse BER performance relative to modulations recommended by the downselection process PCM/PM/NRZ TBD

Gaussian (BT_b = 0.5) Precoded GMSK (h = 0.5) - Req'd Eb/No at 1E-5 BER <= 11.0 dB Gaussian (BT_b = 0.25) - Req'd bandwidth (considering all potential applicable GMSK (h = 0.5) Gaussian (BT_b = 0.3) codes) <= 6 MHz (standard S-band allocation) - Spectral efficiency >= 0.95 using 99% bw and SRRC (roll-off factor = 1.0) assuming a rate 1/2 code (i.e., meet the NTIA out-of- S-band OQPSK (SQPSK) SRRC (roll-off factor = 0.5) band emission mask) - Medium or high hardware maturity required (i.e., keep SRRC (roll-off factor = 0.2) return link relatively simple and low risk) Return/Downlink ≤ 60 Butterworth 6th order - Special exceptions to this process have been made for OQPSK/PM modulations which help ensure carr acq at low C/No, Bessel 6th order i.e., PCM/PSK/PM and PCM/PM/NRZ FQPSK-B Defined by modulation - Although not recommended here, differential PSK modulations should be considered in cases where a low- Defined by modulation SOQPSK-A complexity, high-reliability link design is required - this SOQPSK-B Defined by modulation safety is at the expense of worse BER performance relative to modulations recommended by the PCM/PSK/PM TBD downselection process PCM/PM/NRZ TBD

Precoded GMSK (h = 0.5) Gaussian (BT_b = 0.5)

- Req'd Eb/No at 1E-5 BER <= 11.0 dB SRRC (roll-off factor = 1.0) - Req'd bandwidth (considering all potential applicable OQPSK (SQPSK) codes) <= 6 MHz (standard S-band allocation) - Spectral efficiency >= 0.95 using 99% bw (i.e., meet Forward/Uplink > 60 SRRC (roll-off factor = 0.5) the NTIA out-of-band emission mask) - High hardware maturity required (i.e., keep forward Butterworth 6th order contigency link simple and low risk; no trellis receiver possible) OQPSK/PM Bessel 6th order

Table 11-5: Category A Constellation Link Modulation Initial Down-select Re- sults (also applicable to the Mars Relay Scenario)

CMLP Study Page 133 August 2007 CMLP Final Report

Gaussian (BT_b = 0.5) Precoded GMSK (h = 0.5) Gaussian (BT_b = 0.25) GMSK (h = 0.5) Gaussian (BT_b = 0.3) SRRC (roll-off factor = 1.0) - Req'd Eb/No at 1E-5 BER <= 11.0 dB - Req'd bandwidth (considering all potential applicable OQPSK (SQPSK) SRRC (roll-off factor = 0.5) codes) <= 6 MHz (standard S-band allocation) Return/Downlink > 60 SRRC (roll-off factor = 0.2) - Spectral efficiency >= 0.95 using 99% bw and assuming a rate 1/2 code (i.e., meet the NTIA out-of- Butterworth 6th order OQPSK/PM band emission mask) Bessel 6th order - Medium or high hardware maturity required

FQPSK-B Defined by modulation SOQPSK-A Defined by modulation SOQPSK-B Defined by modulation

SRRC (roll-off factor = 1.0) - Req'd Eb/No at 1E-5 BER <= 11.0 dB - Req'd bandwidth (considering all potential applicable OQPSK (SQPSK) codes) <= 6 MHz (standard S-band allocation) - Spectral efficiency >= 1.14 using 99% bw and Forward/Uplink 6,000 SRRC (roll-off factor = 0.5) assuming a rate 7/8 code (must fit in 6 MHz bandwidth allocation) S-band - High hardware maturity required (i.e., keep forward contigency link simple and low risk; no trellis receiver OQPSK/PM Butterworth 6th order possible)

Precoded GMSK (h = 0.5) Gaussian (BT_b = 0.25) SRRC (roll-off factor = 1.0) - Req'd Eb/No at 1E-5 BER <= 11.0 dB OQPSK (SQPSK) SRRC (roll-off factor = 0.5) - Req'd bandwidth (considering all potential applicable SRRC (roll-off factor = 0.2) codes) <= 6 MHz (standard S-band allocation) Return/Downlink 6,000 - Spectral efficiency >= 1.14 using 99% bw and OQPSK/PM Butterworth 6th order assuming a rate 7/8 code (must fit in 6 MHz bandwidth allocation) FQPSK-B Defined by modulation - Medium or high hardware maturity required SOQPSK-A Defined by modulation SOQPSK-B Defined by modulation - Req'd bandwidth (considering all potential applicable SRRC (roll-off factor = 0.5) codes) <= 10 MHz ; this basically mandates a spectral 8PSK efficiency >= 2.3 using 99% bw and assuming a rate 7/8 code Downlink 20,000 SRRC (roll-off factor = 0.35) - Req'd Eb/No at 1E-5 BER <= 14.0 dB (TBD, need modulation catalog updated to state correct filtered modulation performance) 16-APSK (12,4) SRRC (roll-off factor = 0.5) - Peak-to-Average Power Ratio < 5 dB Table 11-5: Category A Constellation Link Modulation Initial Down-select Re- sults (also applicable to the Mars Relay Scenario) (Continued 1)

CMLP Study Page 134 August 2007 CMLP Final Report

Gaussian (BT_b = 0.5) Precoded GMSK (h = 0.5) Gaussian (BT_b = 0.25) GMSK (h = 0.5) Gaussian (BT_b = 0.3) SRRC (roll-off factor = 1.0) - Req'd bandwidth (considering all potential applicable OQPSK (SQPSK) SRRC (roll-off factor = 0.5) codes) <= 50 MHz (maximum Ka-band allocation); this basically mandates a spectral efficiency >= 0.95 using Forwardlink 6,000 SRRC (roll-off factor = 0.2) 99% bw and assuming a rate 7/8 code Butterworth 6th order - Req'd Eb/No at 1E-5 BER <= 11.0 dB OQPSK/PM - Medium or high hardware maturity required Bessel 6th order

FQPSK-B Defined by modulation SOQPSK-A Defined by modulation SOQPSK-B Defined by modulation Ka-Band Gaussian (BT_b = 0.5) Precoded GMSK (h = 0.5) Gaussian (BT_b = 0.25) GMSK (h = 0.5) Gaussian (BT_b = 0.3)

SRRC (roll-off factor = 1.0) - Req'd bandwidth (considering all potential applicable OQPSK (SQPSK) SRRC (roll-off factor = 0.5) codes) <= 300 MHz (maximum Ka-band allocation); this basically mandates a spectral efficiency >= 0.95 using Returnlink 25,000 SRRC (roll-off factor = 0.2) 99% bw and assuming a rate 7/8 code Butterworth 6th order - Req'd Eb/No at 1E-5 BER <= 11.0 dB OQPSK/PM - Medium or high hardware maturity required Bessel 6th order

FQPSK-B Defined by modulation SOQPSK-A Defined by modulation SOQPSK-B Defined by modulation Table 11-5: Category A Constellation Link Modulation Initial Down-select Re- sults (also applicable to the Mars Relay Scenario) (Continued 2)

CMLP Study Page 135 August 2007 CMLP Final Report

Remaining Modulations after Link Description First Stage Downselect(1,2) First Stage Downselect Process Link Name (Eliminate modulations which underperform on certain critical FOMs) Modulation Shaping/Filter Link Type Symbol Rate ID Type

FSK Meet the NTIA mask

Differential PSK Meet the NTIA mask - High hardware maturity required (i.e., keep forward link Forward Low Rate simple and low risk; no trellis receiver possible) PCM/PSK/PM Meet the NTIA mask - Must include a residual carrier to aid acquisition

PCM/PM/NRZ or Bi-phase Meet the NTIA mask

Precoded GMSK (h = 0.5) Gaussian (BT_b = 0.5)

SRRC (roll-off factor = 1.0) OQPSK (SQPSK) - Req'd Eb/No at 1E-5 BER <= 11.0 dB Forward High Rate SRRC (roll-off factor = 0.5) - High hardware maturity required (i.e., keep forward link simple; no trellis receiver possible) Butterworth 6th order OQPSK/PM Bessel 6th order

FSK Meet the NTIA mask

Differential PSK Meet the NTIA mask - Medium or high hardware maturity required (i.e., keep < 180 ksps Mars, else < 360 return link relatively simple and low risk) ksps PCM/PSK/PM Meet the NTIA mask - Must include a residual carrier to aid acquisition

PCM/PM/NRZ or Bi-phase Meet the NTIA mask

Gaussian (BT_b = 0.5) Precoded GMSK (h = 0.5) Gaussian (BT_b = 0.25)

GMSK (h = 0.5) Gaussian (BT_b = 0.3) X-band SRRC (roll-off factor = 1.0) - Req'd bandwidth <= 8.0 MHz for Mars mission and <= 12.0 OQPSK (SQPSK) SRRC (roll-off factor = 0.5) MHz for Non-Mars Mission - Spectral efficiency >= 0.75 using 25 dB bw < 6 Msps Mars, else < 9 Msps SRRC (roll-off factor = 0.2) - Req'd Eb/No at 1E-5 BER <= 11.0 dB Butterworth 6th order - Medium or high hardware maturity required (i.e., keep OQPSK/PM return link relatively simple and low risk) Bessel 6th order Return FQPSK-B Defined by modulation

SOQPSK-A Defined by modulation

SOQPSK-B Defined by modulation

8PSK SRRC (roll-off factor = 0.35)

SRRC (roll-off factor = 0.5) - Req'd bandwidth <= 8.0 MHz for Mars mission and <= 12.0 16-APSK (12,4) MHz for Non-Mars Mission < 18 Msps Mars, else < 27 SRRC (roll-off factor = 0.35) - Spectral efficiency >= 2.25 using 25 dB bw Msps - Req'd Eb/No at 1E-5 BER <= 14.0 dB SRRC (roll-off factor = 0.5) - Peak-to-Average Power Ratio < 6.25 dB 16-QAM SRRC (roll-off factor = 0.35)

16-APSK (12,4) SRRC (roll-off factor = 0.35) - Req'd bandwidth <= 8.0 MHz for Mars mission - Spectral efficiency >= 3.0 using 25 dB bw < 24 Msps Mars - Req'd Eb/No at 1E-5 BER <= 14.0 dB 16-QAM SRRC (roll-off factor = 0.35) - Peak-to-Average Power Ratio < 6.25 dB

Precoded GMSK (h = 0.5) Gaussian (BT_b = 0.25) SRRC (roll-off factor = 1.0) - Spectral efficiency >= 0.75 using 25 dB bw OQPSK (SQPSK) - Req'd Eb/No at 1E-5 BER <= 11.0 dB Forward All rates SRRC (roll-off factor = 0.5) -High hardware maturity required (i.e., keep forward link Butterworth 6th order simple; no trellis receiver possible) OQPSK/PM Bessel 6th order Gaussian (BT_b = 0.5) Precoded GMSK (h = 0.5) Gaussian (BT_b = 0.25) GMSK (h = 0.5) Gaussian (BT_b = 0.3) Ka-band SRRC (roll-off factor = 1.0) OQPSK (SQPSK) SRRC (roll-off factor = 0.5) - Spectral efficiency >= 0.75 using 25 dB bw Return All rates SRRC (roll-off factor = 0.2) - Req'd Eb/No at 1E-5 BER <= 11.0 dB - Medium or high hardware maturity required Butterworth 6th order OQPSK/PM Bessel 6th order FQPSK-B Defined by modulation SOQPSK-A Defined by modulation SOQPSK-B Defined by modulation Table 11-6: Category B Link Modulation Initial Down-select Results

CMLP Study Page 136 August 2007 CMLP Final Report

Table 11-7: Category A SN and GN Link Modulation Final Down-select Results

137 CMLP Final Report

Table 11-7: Category A SN and GN Link Modulation Final Down-select Results (Continued 1)

138 CMLP Final Report

139 CMLP Final Report

Table 11-7: Category A SN and GN Link Modulation Final Down-select Results (Continued 2)

140 CMLP Final Report

Table 11-8: Category A Constellation Link Modulation Final Down-select Results

141 CMLP Final Report

142 CMLP Final Report

Table 11-8: Category A Constellation Link Modulation Final Down-select Results (Continued 1)

143 CMLP Final Report

144 CMLP Final Report

Table 11-8: Category A Constellation Link Modulation Final Down-select Results (Continued 3)

145 CMLP Final Report

Table 11-9: Category B Link Modulation Final Down-select Results

146 CMLP Final Report

11.4 Codes

The purpose of coding is to reduce the power needed in order to achieve a given error rate. As such, power efficiency is generally the dominant FOM in the comparison of the various codes. However, any of the other FOMs (spectral efficiency, latency, user burden, etc.) could prevent the use of the most power efficient codes. For example, a link with strict latency requirements can disallow the use of any code with a very large blocklength, because the time to receive and decode a long block exceeds the allowable latency. Or, the available bandwidth of a link may restrict the code rates that may be used, because for a constant data rate lower rate codes use inversely proportionally more spectrum.

The code down-select procedure is performed after the final modulation selection process has completed. In particular, for each link under consideration, we assume the use of the recommended modulation. The spectral efficiency of this modulation allows us, then, to compute the eligible code rates, as we describe below.

11.4.1 Constraints Relevant to Code Selection

The bandwidth assignments (allocations) for various near Earth and deep space bands are given in Table 11-10.

Band Application Bandwidth* Assignment (allocation)

S-band Forward or return 6 MHz S-band Launch 10 MHz X-Band Near Earth forward 150 Mhz X-band Deep space, non-efficient 4 MHz modulations X-band Deep space, with efficient 50 MHz modulation Ka-band Near Earth return 650 MHz Ka-band Deep Space 500 MHz *The bandwidth is measured by SFCG conventions as the 99% bandwidth metric for near Earth, and the 25 dB down metric for deep space. Table 11-10: Bandwidth assignments

The latency requirements are given in Table 11-11.

147 CMLP Final Report

Link Type Application Decoder Latency Requirement Voice Near Earth 100 ms Voice Lunar 250 ms Non-voice Any N/A Table 11-11: Latency requirements

11.4.2 Down-select Procedure

These blocklength and bandwidth restrictions suggest an initial down-select process that eliminates codes from consideration that would violate those restrictions:

1. Compute bandwidth used by recommended modulation at specified data rate, uncoded. For a link using a given data rate R b/s and modulation with spectral efficiency  b/s/Hz, uncoded transmission uses a bandwidth of B = R/ Hz.

2. Compute minimum code rate available, using step 1, and total bandwidth available. The transmission will meet a given bandwidth assignment (allocation) Ba Hz only if the code rate satisfies r ≥ B/Ba. The values of Ba used by the study are given in Table 11-10.

3. Compute maximum input block size k, using latency requirement. The decoding latency is the difference between the time a bit is decoded and the time it's encoded version first begins arriving at the receiver. For a block code, this includes the time for a whole codeblock to arrive at the receiver plus the time it takes to decode it. For a convolutional code, it is the time for a number of bits to arrive that is equal to the traceback depth of the Viterbi decoder plus the time to perform a traceback operation. Given the high-speed decoders that exist today, the study assumed that the latency is dominated by the time it takes to receive the relevant bits to decode.

In order for a block code to meet a latency constraint on a link using a given data rate R b/s and having a latency requirement Tl s, the input block size k of the block code must satisfy k ≤ Tl*R. The values of Tl are given in Table 11-11.

4. Sort code catalog shown in Table 7-2 based on rate r. Eliminate those with disallowable values of r.

148 CMLP Final Report

5. Sort code catalog by input block size k, among codes with eligible rates. Eliminate those with disallowable values of k.

6. Identify top performing code(s) within (k,r) constraints based on performance, complexity, and maturity.

7. Narrow selections using, FOM analysis, to a small set of candidate codes that work for all links.

This procedure may be carried out on each of the links identified in Table 11-7, Table 11-8, and Table 11-9. Although these links are already quite numerous, we found it necessary to further partition the links, by data rate, in order to assure the capture of all latency and bandwidth constraints. For example, in Table 11-7 the first link listed is an operational forward S-band link with a data rate of “≤ 60 kbps.” If it were ex- actly 60 kbps, then the latency requirement in step 3 would require that the blocklength satisfy k ≤ Tl*R = 6000. However, an 18 kbps link would also fall into the “≤ 60 kbps” category, but in that case the blocklength constraint is the more stringent k ≤ Tl*R = 1800. The two cases are sufficiently different that different coding solutions would be recommended. To ensure that different data-rate-dependent link drivers were captured, we partitioned the S-band data rates into ranges: 18 – 100 kbps, 100 – 300 kbps, 300 – 4800 kbps, and 4.8 – 6 Mbps.

11.4.3 Initial Code Down-selections

The initial down-selection involves following the first six steps above. This resulted in the first-stage down-select codes shown in Table 11-12, Table 11-13, Table 11-14. These are the codes, among those that meet the bandwidth and latency requirements that have the highest power efficiency and acceptable complexity and maturity.

149 CMLP Final Report

Table 11-12: First stage code recommendations, near Earth, forward links

150 CMLP Final Report

Link constraints, using recommended Link Description Recommended Modulation(1,2) First Stage Recommended Codes(1) modulation k upper Spectral Input Required Used BW bound(6) Bandwidt Data Rate Available efficiency r lower Code Code block Latency Eb/No to Direction Band Modulation ID Shaping/Filter Type (Mhz), based on h used (Mbps) BW (Mhz) (b/s/Hz), bound(5) ID rate (r) length (msec) achieve uncoded(4) 100 msec (Mhz) 99% BW(3) (k) BER=1e-8 latency Min Max Turbo(1784,1/6) 0.166 1784 99 0.5 0.57 0.018 0.1 0.086206897 0.014 1800 TPC(16,11)x(16,11)x(16,11) 0.325 1331 74 0.3 1.80 AR4JA(1024,1/2) 0.500 1024 57 0.2 2.17 Turbo(8920,1/2) 0.500 8920 89 5.2 1.3 0.1 3 2.586206897 0.431 10000 AR4JA(4096,1/2) 0.5 4096 41 5.2 1.39 Precoded GMSK Gaussian (BT_b = 1.16 TPC(H64xH32xS32) 0.701 45942 15 5.9 2.50 (h = 0.5) 0.25) 3 4.8 4.137931034 0.690 300000 AR4JA(16384,4/5) 0.800 16384 55.22.82 AR4JA(4096,4/5) 0.800 4096 15.23.21 TPC(128,120)^2 0.879 14400 35.93.90 4.8 6 5.172413793 0.862 480000 F-LDPC(16k, 8/9) 0.889 16384 35.83.97 C2, 50 iterations 0.875 7136 15.94.07 Turbo(1784,1/6) 0.166 1784 99 0.4 0.57 0.018 0.1 0.063291139 0.011 1800 TPC(16,11)x(16,11)x(16,11) 0.325 1331 74 0.2 1.80 AR4JA(1024,1/2) 0.500 1024 57 0.1 2.17 Turbo(8920,1/3) 0.333 8920 89 5.7 0.58 0.1 3 1.898734177 0.316 10000 AR4JA(4096,1/2) 0.500 4096 41 3.8 1.39 OQPSK SRRC (roll-off factor 6 1.58 AR4JA(16384,1/2) 0.500 16384 56.11.03 (SQPSK) = 0.5) S-band 3 4.8 3.037974684 0.506 300000 Turbo(8920,1/2) 0.500 8920 36.11.30 AR4JA(4096,1/2) 0.500 4096 16.11.39 AR4JA(16384,2/3) 0.667 16384 35.71.85 4.8 6 3.797468354 0.633 480000 AR4JA(4096,2/3) 0.667 4096 15.72.20 F-LDPC(16k, 2/3) 0.667 16384 35.72.11 Turbo(1784,1/6) 0.166 1784 99 0.5 0.57 0.018 0.1 0.087719298 0.015 1800 TPC(16,11)x(16,11)x(16,11) 0.325 1331 74 0.3 1.80 AR4JA(1024,1/2) 0.500 1024 57 0.2 2.17 Turbo(8920,1/2) 0.500 8920 89 5.3 1.3 0.1 3 2.631578947 0.439 10000 AR4JA(4096,1/2) 0.5 4096 41 5.3 1.39 OQPSK/PM Butterworth 6th order 1.14 TPC(H64xH32xS32) 0.701 45942 15 6.0 2.50 3 4.8 4.210526316 0.702 300000 AR4JA(16384,4/5) 0.800 16384 55.32.82 AR4JA(4096,4/5) 0.800 4096 15.33.21 TPC(128,120)^2 0.879 14400 36.03.90 4.8 6 5.263157895 0.877 480000 F-LDPC(16k, 8/9) 0.889 16384 35.93.97 C2, 50 iterations 0.875 7136 16.04.07 SRRC (roll-off factor Uncoded 11 0 8.6 15.4 16 22 8PSK 10 2.57 8.560311284 0.856 1600000 Return = 0.35) C2, 50 iterations 0.875 7136 09.84.07 150 Turbo(8920,1/3) 0.333 8920 9 129.4 0.58 150 43.10344828 0.287 100000 150 AR4JA(16384,1/2) 0.500 16384 16 86.2 1.03 150 AR4JA(16384,2/3) 0.667 16384 0 129.3 1.85 50 100Precoded GMSK Gaussian (BT_b = 86.20689655 0.575 5000000 150 1.16 TPC(S16xH32^2) 0.619 10140 0 139.3 2.42 (h = 0.5) 0.25) 150 TPC(128,120)^2 0.879 14400 0 147.1 3.90 100 150150 129.3103448 0.862 10000000 F-LDPC(16k, 8/9) 0.889 16384 0 145.5 3.97 150 C2, 50 iterations 0.875 7136 0 147.9 4.07 150 Turbo(8920,1/3) 0.333 8920 9 131.6 0.58 150 43.85964912 0.292 100000 150 AR4JA(16384,1/2) 0.500 16384 16 87.7 1.03 150 AR4JA(16384,2/3) 0.667 16384 0 131.6 1.85 50 100 87.71929825 0.585 5000000 OQPSK/PM Butterworth 6th order 150 1.14 TPC(S16xH32^2) 0.619 10140 0 141.7 2.42 150 TPC(128,120)^2 0.879 14400 0 149.7 3.90 100 150 150 131.5789474 0.877 10000000 F-LDPC(16k, 8/9) 0.889 16384 0 148.0 3.97 X-band 150 C2, 50 iterations 0.875 7136 0 150.5 4.07 150 Turbo(8920,1/4) 0.250 8920 9 126.6 0.27 150 31.64556962 0.211 100000 150 AR4JA(16384,1/2) 0.500 16384 16 63.3 1.03 150 AR4JA(16384,1/2) 0.500 16384 0 126.6 1.03 50 100 OQPSK SRRC (roll-off factor 63.29113924 0.422 5000000 150 1.58 AR4JA(16384,2/3) 0.667 16384 094.91.85 (SQPSK) = 0.5) 150 AR4JA(16384,2/3) 0.667 16384 0 142.4 1.85 100 150 150 94.93670886 0.633 10000000 F-LDPC(16k, 2/3) 0.667 16384 0 142.4 2.11 150 TPC(H64xH32xS32) 0.701 45942 0 135.4 2.50 150 TPC(128,120)^2 0.879 14400 0 144.0 3.90 SRRC (roll-off factor 150 2.37 126.5822785 0.844 F-LDPC(16k, 8/9) 0.889 16384 0 142.4 3.97 = 0.5) 150 C2, 50 iterations 0.875 7136 0 144.7 4.07 240 300 8PSK 24000000 150 TPC(128,120)^2 0.879 14400 0 132.8 3.90 SRRC (roll-off factor 150 2.57 116.7315175 0.778 F-LDPC(16k, 8/9) 0.889 16384 0 131.3 3.97 = 0.35) 150 C2, 50 iterations 0.875 7136 0 133.5 4.07 Precoded GMSK Gaussian (BT_b = 1.16 560.3448276 0.862 100000 7 640.8 (h = 0.5) 0.25) C2, 50 iterations 0.875 7136 4.07 1650 OQPSK SRRC (roll-off factor 1.58 411.3924051 0.633 100000 16 617.1 Ka-band (SQPSK) = 0.5) 650 AR4JA(16384,2/3) 0.667 16384 1.85 OQPSK/PM Butterworth 6th order 1.14 570.1754386 0.877 100000 C2, 50 iterations 0.875 7136 7 652.0 4.07 OQPSK SRRC (roll-off factor 650 1000 591.7159763 0.910 65000000 0 676.6 (SQPSK) = 0.2) 1.69 C2, 50 iterations 0.875 7136 4.07 Table 11-13: First stage code recommendations, near Earth, return links

151 CMLP Final Report

Table 11-14: First stage code recommendations, deep space

152 CMLP Final Report

As can be seen, a relatively small set of code candidates covered all links:

 Legacy codes: uncoded, (7,1/2) convolutional, and BCH (63,56)

 AR4JA & C2 – nearly the entire family of CCSDS orange book codes o (1024,1/2), (4096, {1/2, 2/3, 4/5}), (16384, {1/2, 2/3, 4/5}), C2

 Turbo – CCSDS blue book codes of longest and shortest lengths o (1784, {1/4, 1/6}), (8920, {1/2, 1/3, 1/4})

 Turbo Product codes – 2D and 3D versions o TPC(128,120)2, TPC(16,11)3, TPC(H64xH32xS32)

 F-LDPC o (16k, 2/3), (16k, 8/9)

11.4.4 FOM Analysis For each link, we ranked the codes surviving the first stage selection process by each of the ten FOMs. For each link, the FOMs are weighted to arrive at a final FOM score. A couple of observations are in order regarding these weights. First, because the initial down-select procedure weeded out those codes that did not meet the bandwidth constraint, the remaining codes all meet the bandwidth constraint, and so there may be limited value in preferring one code over another with respect to this FOM. This is true even when the bandwidth constraint is very important or stringent. For example, if the bandwidth constraint forces the code rate to be 7/8 or higher, then the vast majority of codes from the catalog are eliminated from consideration in the initial down-select process, but those that are remaining are not preferred over one another on the bases of spectral efficiency because all remaining codes meet the constraint. Therefore, the weighting of spectral efficiency (and latency) are quite low in the final FOM ranking and analysis.

A second FOM consideration is that the links do not use the same FOM weightings. This is a consequence of different mission screnarios giving rise to different priori- ties. For example, an uplink for an outer planets mission would weight the power efficiency FOM higher than a LEO mission, because the outer planet mission may have a much harder struggle to meet its data rate requirements without building significantly enhanced ground infrastructure, compared to a LEO mission.

We now describe how the ranking was done for each of the ten FOMs.

1. Supports legacy missions

153 CMLP Final Report

The highest rank was assigned to codes that are currently flying on missions. The next highest rank was given to codes that plan to use the code, and the lowest rank was given to all other codes.

2. Spectral utilization

The ranking of spectral utilization was based on the entry in the table corresponding to “bandwidth used,” which itself is a function of the given data rate and modulation spectral efficiency, and the rate of the code under consideration. Therefore, the ranking of the codes is a ranking of the code rates from highest to lowest. Since the first stage down-select process has already yielded a set of candidate codes of approximately the same code rate, there is no need to give additional weighting to spectral utilization in the final FOM analysis.

3. Power efficiency

The ranking of power efficiency was based on the required Eb/N0 needed in order to achieve BER = 10-8. This is the error rate requirement of the Constel- lation missions, but is otherwise arbitrary. Except for codes with known error floors, most notably the turbo codes, the particular choice of error rate requirement does not substantially affect the power efficiency ranking of the codes. This is because the codes surviving the first stage down-select process are of roughly the same rate and length and are top-performing-- thus, they have approximately the same slope in the waterfall region.

4. User burden

This is a measure of the cost to a mission of using a particular code. Codes of equivalent flight heritage – whether extensive or nonexistent – were assigned the same ranking.

5. Infrastructure burden

Codes already supported by the infrastructure (SN, GN, and DSN) were assigned higher ranking than those not supported by the infrastructure. Among those not already supported, the relative ranking of codes reflects the anticipated cost to implement their support.

6. Alignment with international standards

Codes in the CCSDS Blue Book were assigned the highest ranking. CCSDS Orange Book codes were assigned the next highest ranking. Codes that are

154 CMLP Final Report

IEEE, ITU, DVB or other standards for non-space applications were assigned the next highest ranking. The last ranking was used for codes that are not known to be part of any standard.

7. Robustness

With respect to coding, “robustness” captures (a) the ability of the code to operate in the presence of carrier synchronization error, symbol timing error, and non-AWGN noise, (b) the ability to detect when a decoder is unable to decode correctly (as opposed to putting out a decoded stream without knowing whether it is in error), and (c) the lack of an error floor at BER < 10-8. As a general guide, codes surviving the first stage down-select and having similar code rate and length have been observed to have similar performance with respect to (a). Therefore, the ranking was based primarily on (b) and (c).

8. Latency

As mentioned above, the first stage down-select process has resulted in codes that are approximately the same length, and therefore, latency is given little additional weight in the final FOM analysis. For block codes, the ranking of latency is based on the length of the block code. For the convolutional code, the ranking is based on the length of the traceback.

9. TRL

Codes with flight heritage are ranked the highest. Those with space technology demonstrations are assigned the next highest rank. Those with planned technology demonstrations are next, followed by those without any plans for space flight, and last, those without any known hardware implementation in the laboratory.

10. Capacity

For the purposes of comparison, the study assumes use of an AWGN channel, which is a good approximation for communications from space. The ranking of aggregate capacity, then, is a function of the Eb/N0 required of the number of simultaneous links that are supported. Since the first stage down-select procedure has resulted in codes of roughly the same code rate, the number of simultaneous links is the same for each candidate code, and the ranking reduces to a duplication of the power efficiency measure. As such, this FOM is given a low weight in the scoring system.

155 CMLP Final Report

11.4.5 Final Code Selections The initial code selections (first stage down-select) identified the top codes for each link. In each case, we identified up to three candidate codes. For each link, we ranked the candidate codes using the FOM analysis of the previous section, and computed a weighted-average FOM rank.

Following step 7 of Section 11.4.2, we next identified the smallest set of recommended codes that would work well for all links scenarios. The FOM analysis indicates that for all links, CCSDS turbo and AR4JA LDPC codes uniformly outranked the turbo product codes and Flarion LDPC codes, and so they were eliminated. Among the remaining codes there was no smaller subset of codes that uniformly outperformed another code. This left uncoded, convolutional, turbo, AR4JA LDPC, and C2 LDPC codes. Each of these codes had the best FOM score for at least one reference link, and so it would not be possible to reduce the set of recommended codes further without sacrificing link performance.

Table 11-15 provides the recommended codes for Category A near-Earth SN and GN links. Table 11-16 provides the recommended codes for Category B links. Although “International Standardization” is only one FOM, the final recommended links are in happy accordance with the CCSDS standards and ongoing working group activities. All the recommended codes are either contained in the CCSDS Blue Book (convolutional, turbo) or in an experimental CCSDS Orange Book (AR4JA LDPC, C2 LDPC), as indicated in the color-coding of the tables.

156 CMLP Final Report

Link Description Recommended Codes Input Comments Direction Band BW (MHz) Data Rate (Mbps) Code ID Rate length CC offers best latency Š use when < 0.001 CC(7,1/2) < 1000 realtime operation needed at < 1 kbps 1024 to Best coding gain; lower complexity and 0.001 to 3 AR4JA LDPC 16384 error floor than r=1/2 turbo S-band 6 Forward 1024 to 3 to 4.8 AR4JA LDPC 2/3, 4/5 (Uplink) 16384 High bandwidth efficiency; better coding > 4.8 C2 LDPC 0.87 7136 gain than RS-only 1024 to X-band 50 < 25 AR4JA LDPC 16384

< 0.001 CC(7,1/2) < 1000 1024 to 0.001 to 3 AR4JA LDPC 16384 S-band 6 1024 to 3 to 4.8 AR4JA LDPC 2/3, 4/5 16384

> 4.8 C2 LDPC 0.87 7136 S-band 1024 to Return 20 16 to 22 AR4JA LDPC (launch) 16384 (Downlink) 1/6, 1/4, < 50 Turbo 8920 1/3, 1/2 X-band 50 1024 to 50 to 150 AR4JA & C2 LDPC 0.5 to 0.87 16384 1/6, 1/4, < 300 Turbo 8920 1/3, 1/2 Ka-band 650 1024 to 300 to 650 AR4JA & C2 LDPC 1/2 to .87 16384 Blue = CCSDS Blue Book Standard Orange = CCSDS Orange Book Specification Table 11-15: Category A Code Recommendations

Link Description Recommended Codes Input Comments Direction Band BW (MHz) Data Rate (Mbps) Code ID Rate length CC offers best latency Š use when realtime < 0.001 CC(7,1/2) < 1000 operation needed at < 1 kbps 1024 to Best coding gain at r=1/2; lower complexity 0.001 to 40 AR4JA LDPC 1/2, 2/3, 4/5 16384 & error floor than r=1/2 turbo X-band 50 1784 to 0.001 to 15 Turbo 1/6, 1/4, 1/3 Can use when additional coding gain Forward 8920 needed, and UER not an issue (Uplink) High bandwidth efficiency; better coding > 40 C2 LDPC 0.87 7136 gain than RS-only 1024 to AR4JA LDPC 16384 Ka-band 500 All 1784 to Turbo 1/6, 1/4, 1/3 Can use when additional coding gain 8920 needed, and UER not an issue 1/6, 1/4, < 50 Turbo 8920 1/3, 1/2 X-band 50 AR4JA & C2 1024 to 50 to 150 0.5 to 0.87 Return LDPC 16384 (Downlink) 1/6, 1/4, < 300 Turbo 8920 1/3, 1/2 Ka-band 500 AR4JA & C2 1024 to 300 to 500 1/2 to .87 LDPC 16384 Blue = CCSDS Blue Book Standard Orange = CCSDS Orange Book Specification Table 11-16: Category B Code Recommendations

157 CMLP Final Report

11.5 Multiple Access

Mission scenarios for Near Earth, Lunar, and Mars envision cases where simultaneous communication between multiple users in a single antenna beam is required. Consequently Multiple Access (MA) techniques are needed to satisfy the needs of the communities of these user platforms. Beyond the ability to accommodate simultaneous communications by a set of users, the necessity for simultaneous ranging for most of the user platforms and the need for bandwidth efficiency are also strong drivers in considering which MA techniques to choose. Ranging capability and bandwidth efficiency are among the FOMs in the MA selection process. Other FOMs such as user and infrastructure burden and alignment with international standards also affect the choice of MA scheme.

The Near Earth communication networks that were analyzed for MA included the Earth-based Antenna Elements and the Earth-based Relay Elements. These are referred to below as the Ground Network (GN) and the Space Network (SN), respectively.

11.5.1 Down-select Procedure The study process for Multiple Access follows the outline given in Section 4.5 of this report, specialized to any unique needs. The primary feature unique to the MA study is the necessity to consider jointly the communication and navigation aspects of the techniques.

The MA down-select process is as follows:

1. In preparation for the MA down-select process, a catalog of MA techniques was developed. It is shown in Table 11-17. Also, representative catalogs of missions were developed for Near Earth, Lunar, and Mars scenarios. From these catalogs multiple access scenarios were constructed where requirements for simultaneous communications by multiple users in a single antenna beam were found.

2. By performing an initial down-select based upon the performance of each candidate modulation in the FOM areas of greatest weighting, a large number of MA techniques were eliminated from further consideration. The reduced catalog was submitted to further analysis for the Near Earth, Lunar, and Mars scenarios.

3. Final down-selection for the Near-Earth scenarios, which include supporting MA communications through the Space Network (SN) and Ground Network (GN), proceeded through inspection, analysis, and logic, accounting for the

158 CMLP Final Report

strong legacy of both the SN and GN systems and the unique aspects of providing communications with these systems. Indeed it might be said that the choice of a future MA technique is partially driven by the existence of MA schemes that have been and are still used by the SN and GN to successfully service populations of user platforms in the Near Earth environment

4. Distinct and more intensive analyses were performed on the leading MA candidates for the lunar S-band scenarios - CDMA, CDMA/FDMA, and GMSK/PN - to provide inputs to the FOM analysis. In the process, more specific implementations that would be adequate to service user platforms were articulated. This was done by first developing realistic stressing operational scenarios for a set of user platforms in the lunar environment (for the LRS and the DTE/DFE lunar cases). Capacity analyses were then performed for this operational scenario using theoretical equations for CDMA and GMSK/PN to determine capacity, required bandwidth, and required link power to overcome interference.

5. A formal FOM comparison was performed for the lunar scenarios. For each link, we ranked the MA techniques surviving the first stage selection process by each of the twelve FOMs. The FOM analysis incorporated results from the performance analyses described above, as well as inputs based on engineering judgment and experience.

6. The Mars scenarios for relay orbiter links are expected to be similar to the lunar communications scenarios, and, to a first approximation, we expect Mars relay MA recommendations to parallel lunar MA recommendations, though the FOMs still need to be weighted and evaluated independently. Mars DTE/DFE links are far more power-constrained and are less spectrum- constrained than comparable lunar links, so the FOM weights for Mars DTE/DFE links are significantly different than for lunar links.

7. Final MA recommendations were compiled for each scenario and each link type.

159 CMLP Final Report

MA Scheme Identification D escription

Time Shared Schedule-based time sharing approach

D ata f rom f ixed set of ac tive users TDM Time D ivision Multiplexing multiplexed into one data stream which is transmitted over the channel Each user assigned an individual time slot TDMA Time D ivision Multiple Access Each user transmits during assigned time s lot Each user assigned individual fequency General Frequency D ivision Multiple Access slot (channel) Users transmit simultaneously Subcarriers spaced apart at precise FDMA Orthogonal Frequency D ivision frequencies to achieve orthogonality OFDM Multiplexing among signals and excellent spectral efficiency E ntirely reduces down to F D MA at WDMA Wavelength Division Multiple Access optical frequencies A subchannel is assigned only when DAMA D emand Assigned Multiple access traff ic is available Operates by transmitting different signals simultaneously on diff erent SDMA Space D ivision Multiple Access transmit antennas at the s ame frequency and by using multiple receive antennas for decoding Each user assigned individual pseudo- Code Division Multiple Access random code Traditional D irect Sequence Spread Spectrum Each code is largely orthogonal to any DSSS other assigned code All signals transmitted from a common C onstant node Envelope CDMA Chip synchronous U ser carrier frequencies vary in a FHSS Frequency H opped Spread Spectrum pseudorandom fashion within a wideband PN spread users carrier frequencies D irect Sequence/Frequency H opped DS/FHSS hop periodically in a pseudorandom Spread Spectrum fashion. N o sensing to determine whether the Pure ALOH A R andom access data transfer approach channel is busy or not Slotted ALOHA TD MA-based random access method Time scale slotted into units of time CSMA Similar to Pure ALOHA but with user's C arrier Sense Multiple Access capability to listen to the channel before sending data

C arrier Sense Multiple Access with C apable of notifying others of intent to R andom C S MA /C A C ollision Avoidance transmit bef ore actually s ending data Access C arrier Sense Multiple Access with C apable of detecting collisions and CSMA/CD C ollision D etection stopping transmission immediately Multiple Access with C ollision Acknowledgements required to indicate MACA Avoidance data has arrived TD MA-based random access method U sers contend for open time slots Reservation ALOHA based upon ALOH A through a reservation process C ombination of slotted ALOH A and PR MA Packet R eservation Multiple Access TDMA

FDMA/TDMA D ivide wideband spectrum into ( F D MA /T D M on point-to- subchannels which individually support multipoint links) a self-contained TD MA network

D ivide wideband spectrum into Hybrid FDMA/CDMA subchannels which individually support a self-contained C D MA network

Users assigned a time slot (which may TCDMA Time CDMA change) and a unique PN code U sers can hop to a new frequency at TDFH Time D ivision Frequency H opping the start of a new T D MA frame Time H opping-Pulse Position T ime hopping combined with pulse Other TH-PPM Modulation pos ition modulation Table 11-17: List of All Candidate MA Schemes

160 CMLP Final Report

11.5.2 Initial MA Down-selection Some general assumptions were made going into the initial down-selection to guide the “winnowing” process:

 Alternatives having capacity insufficient to meet link/network scenario data rate and simultaneous user requirements will be dropped

 Capacity/efficiency must remain sufficient to meet link/network scenario data rate and simultaneous user requirements in a high latency environment

 Spectral efficiency must be such that anticipated spectral allocations are sufficient to enable full link/network scenario support

 User burden must be low (engineering judgment call)

 Technology maturity must be medium to high

 Initial down-selection process applies to all scenarios

The initial down-selection results for MA are shown in Table 11-18. Some observations regarding the process:

 Detailed FOM analysis was not required for the Near Earth scenarios. Con- siderations of power flux density restrictions quickly eliminated any non- spread MA techniques for the SN case. For the GN case, the legacy of the existing antenna systems largely dictated the recommendations.

 In performing the detailed analysis underpinning the down-selection for the lunar S-band scenario, most time was spent evaluating CDMA, FDMA, and hybrid CDMA/FDMA approaches, due to time constraints and a sense that these schemes have the most to offer in providing MA capability for the lunar environment. To the extent possible, comparable analyses were performed for the GMSK/PN alternatives.

 Random access methods were eliminated in the initial down-selection. They were seen as incompatible with point-to-multipoint forward links, while on the return link they cause large latency. Since lunar surface links weren’t in the scope of the CMLP study, the scenario where the random access methods would have been important wasn’t addressed.

161 CMLP Final Report

 Implementations of the MA schemes surviving the initial down-select were developed in the second down-selection phase, and are described more fully below.

Remaining Link Type Initial Downselect Comments MA Schemes CDMA

FDMA Random access techniques are incompatible with point-to-multipoint links TDM - time-shared approach eliminated due to inability to meet simultaneous users requirement Forward (Point- dictated by link/network scenario FDMA/CDMA to-Multipoint) - DAMA eliminated due to the effect of large latency on network efficiency - TH-PPM eliminated due to the difficulty in establishing and maintaining strict time synchronization FDMA/TDM among the system and users as well as stringent user position accuracy requirements TCDMA

Operational TDFH (S-band, X-band) - Random access techniques eliminated due to the effect of large latency on network efficiency CDMA - TDMA eliminated due to the difficulty in establishing and maintaining strict time synchronization among the system and users - TH-PPM eliminated due to the difficulty in establishing and maintaining strict time synchronization Return among the system and users as well as stringent user position accuracy requirements (Multipoint-to- FDMA - Time-shared approach eliminated due to inability to meet simultaneous users requirement Point) dictated by link/network scenario - DAMA eliminated due to the effect of large latency on network efficiency - Hybrid techniqes FDMA/TDMA, TCDMA and TDFH eliminated due to the difficulty in establishing FDMA/CDMA and maintaining strict time synchronization among the system and users

FDMA - Random access techniques are incompatible with point-to-multipoint links - Time-shared approach eliminated due to inability to meet simultaneous users requirement dictated by link/network scenario Forward (Point- TDM - DAMA eliminated due to the effect of large latency on network efficiency to-Multipoint) - TH-PPM eliminated due to the difficulty in establishing and maintaining strict time synchronization among the system and users as well as stringent user position accuracy requirements FDMA/TDM

High Rate Science - Random access techniques eliminated due to the effect of large latency on network efficiency (Ka-band) - TDMA eliminated due to the difficulty in establishing and maintaining strict time synchronization among the system and users - TH-PPM eliminated due to the difficulty in establishing and maintaining strict time synchronization among the system and users as well as stringent user position accuracy requirements Return - Time-shared approach eliminated due to inability to meet simultaneous users requirement (Multipoint-to- FDMA dictated by link/network scenario Point) - DAMA eliminated due to the effect of large latency on network efficiency - Hybrid techniqes FDMA/TDMA, TCDMA and TDFH eliminated due to the difficulty in establishing and maintaining strict time synchronization among the system and users - CDMA eliminated due to excessive bandwidth expansion on high rate links and expected high hardware complexity for extreme chip rates - FDMA/CDMA eliminated due to expected high hardware complexity

Table 11-18: Multiple Access Down-select Initial Results

11.5.3 Near Earth Scenarios – Final Selection Analysis This section presents the analysis leading to the MA scheme recommendations for the Near Earth scenarios: simultaneous communications within a single antenna beam, with multiple user platforms, in the Near Earth environment using the SN and the GN.

162 CMLP Final Report

11.5.3.1 SN Scenario

Analysis for the SN scenario is presented for S-band and Ku/Ka-band links, and forward and return direction using link traffic projected for these cases as presented in Table 5-1 and Table 5-2. Most of the customers’ communications will not be geo- graphically located in the same area, and therefore cannot be simultaneously serviced via one antenna beam. They can be handled through scheduled services via single access antennas at S-, Ku-, and Ka-band; multiple scheduled services via the MA phased array at S-band (return services through TDRS H,I,J); or by multiple antenna beams via the MA phased array at S-band (return services through TDRS F1- 7) using the Demand Access System (DAS).

Scenarios in which one could envision multiple users transmitting or receiving via a single antenna beam might include simultaneous support of ISS and CEV or CEV and LSAM during rendezvous and docking in LEO. Alternatively, Global Precipita- tion Mission (GPM) support may require simultaneous communications with two user platforms.

11.5.3.1.1 S-band Links

FDMA schemes are ruled out immediately for this scenario by considering the power flux density (PFD) levels on the Earth resulting from satellite transmissions and comparing these to the PFD requirements set forth by the ITU and NTIA. The analyses for both CDMA and FDMA multiple access schemes are shown below:

S-band PFD Calculation: Return Link Parameter CDMA FDMA Comment Transmit Power (dBW) 12.4 12.4 User with omni antenna at 10 kbps User Altitude (km) 400 400 Elev Angle from Surface (deg) 5 5 Earth Radius (km) 6378 6378 Orbital Radius (km) 6778 6778 Theta (deg) 15.4 15.4 Interior Earth angle Slant Range (km) 1804.5 1804.5 From user satellite to point on Earth Ref BW (Hz) 4000 4000 Defined in Table D-2 of SNUG Spread rate for CDMA, 10 kbps with Rate 1/2 code Symbol Rate (Hz) 6000000 20000 for FDMA PFD limit = -154 dBW/m2 as defined in Table D-2 of PFD (dBW/m2) -155.5 -130.7 SNUG for angle of arrival = 0 to 5 deg S-band PFD Calculation: Forward Link Parameter CDMA FDMA Comment Transmit Power (dBW) 42 42 TDRS EIRP on SSL link Slant Range (km) 40582 40582 From TDRS to point on Earth Ref BW (Hz) 4000 4000 Defined in Table D-2 of SNUG Spread rate for CDMA, 10 kbps with Rate 1/2 code Symbol Rate (Hz) 6000000 20000 for FDMA PFD limit = -144 dBW/m2 as defined in Table D-2 of PFD (dBW/m2) -152.9 -128.1 SNUG for angle of arrival = 25 to 90 deg Table 11-19: PFD Calculations for Near-Earth S-band Scenario

163 CMLP Final Report

As can be seen the PFD levels are at an unacceptably high level for FDMA. On this basis FDMA is excluded as a multiple access technique for S-band near-Earth SN communications. Any other narrow band scheme, e.g. TDMA, would experience similar PFD levels and suffer exclusion for the same reason. Given the PFD requirements stated above, the recommended MA scheme will need to include some sort of spread spectrum capability, such as PN/CDMA.

Recommendation: For SN forward or return S-band multiple access, use CDMA with spread spectrum PN codes through a multiple access phased array antenna.

11.5.3.1.2 Ku- and Ka-band Links

The higher data rates required at Ku- and Ka-band make CDMA less applicable as a multiple access scheme. Within the SN, PN codes are available for Ku-band forward and return and for Ka-band forward, but their use is limited to no more than 300 kbps. In the very rare cases when simultaneous communications is required to two co-located users, it makes sense to employ an FDMA scheme. This is possible at the relay satellite level using the SN, for example, where two single access antenna services may be provided simultaneously. Some frequency separation is available for the two Ka-band services, but not for Ku-band which must rely on spatial separation.

Recommendation: For SN forward or return Ku- or Ka-band multiple access, use an FDMA-type scheme using scheduled services through multiple single access antenna beams with unique frequency slots for each user.

11.5.3.2 GN Scenario

For the GN scenario, various ground antennas at different Earth ground locations provide forward S-band support to LEO and GEO customers. It is very unlikely that simultaneous users could be serviced by one antenna beam simultaneously due to the narrowness of the beams, except in the case of LSAM/CEV or CEV/ISS rendezvous. In addition individual GN antennas can generally only service individual LEO users for short periods at a time due to LOS restrictions.

Recommendation: For multiple access use scheduled services through multiple ground antennas which themselves employ multiple frequency bands.

11.5.4 Lunar Scenarios The NASA lunar communications and navigation architecture is currently under development. During 2007, a study was performed by the Lunar Architecture Team,

164 CMLP Final Report designated the LAT2 Study. The LAT2 Study Final Report describes a Lunar Net- work (LN) to support operations in the lunar vicinity.

Direct to and from Earth (DTE/DFE) RF links can provide continuous coverage to most of the surface on the near side of the moon. Many other areas of interest on the lunar surface, including polar sites where a lunar outpost may be sited to support exploration, resource recovery and activities in preparation for human exploration of Mars, are not in view of Earth and require alternate communication paths; the LAT2 Study proposed relay orbiters and a Lunar Communications terminal on the surface. If surface elements are close together, i.e. within line of sight, lunar surface communications and navigation terminals may reduce the burden of communications among surface elements and between the moon and Earth and aid, along with Lunar Relay Satellites (LRSs), precision navigation for landing and ascent of lunar lander vehicles and position location of roving vehicles and EVA astronauts on the surface. The LN architecture proposed by the LAT2 is one potential architecture for sustained exploration and operations on the moon.

The LN consists of Lunar Relay Satellites (LRSs) supporting S-Band communications and tracking and Ka-Band communications with orbiting and lunar surface users The LN also features Lunar Communications Terminals (LCTs) on the surface providing S-Band communications and tracking and Ka-Band communications with orbiting space vehicles, LRSs and ground terminals on Earth. The LCT also provides a high data rate commercial wireless local area network (WLAN) supporting surface communications.

165 CMLP Final Report

Figure 11-1: Lunar Communications Concept The LAT2 Study concept of operations envisions WLAN support for lunar rovers, EVA astronauts and transport service and excavator vehicles within about a six ki- lometer radius of the lunar outpost by an LCT which then communicates with other surface elements, the LRSs or directly to Earth when it’s in view primarily using Ka- Band.

Excursions away from the outpost LCT coverage zone utilize S-Band or Ka-Band on the rover to communicate with the outpost and Earth via the LRS or direct to Earth, if necessary. EVA astronauts communicate with each other, the outpost and Earth via the WLAN to the LCT in the outpost zone or the rover on excursions. S-Band and Ka-Band radios are used by surface elements to the LRS in locations where there are obstructions or gaps in the WLAN coverage.

166 CMLP Final Report

Figure 11-2: Lunar Surface Excursions – Autonomous Position Location Concept

S-Band is used extensively for navigation to provide precision landing and ascent of the lunar lander vehicle and position location of mobile elements on the surface. Multipoint-to-point tracking signals from LRSs and surface LCTs enhance the ability to do autonomous precision navigation of lander vehicles and position location of vehicles or EVA astronauts on the lunar surface.

167 CMLP Final Report

Figure 11-3: LRS/LCT Autonomous Precision Landing Navigation Concept

S-Band DTE/DFE links provide communications and navigation to Orion/LSAM during transit between the Earth and moon and in lunar orbit when not obscured by the moon. DTE/DFE range and Doppler measurements can be used for IMU updates on lunar surface vehicles from range measurements from three widely separated Earth stations. Ranging is not required on links between the moon’s surface and Earth if there is a lunar relay network in place to perform the ranging function

11.5.5 Code Division Multiple Access for Lunar Missions This section addresses three simultaneous user scenarios using CDMA signal designs under different assumptions regarding the lunar communications and navigation architecture and operations concepts that will be in place.

11.5.5.1 CDMA Overview

The CDMA technique analyzed for use in the lunar scenario is the same Space Net- work Interoperability Panel (SNIP)- based design recommended by the CMLP Study for use by the near Earth relay for supporting missions in low Earth orbit.

168 CMLP Final Report

The Direct Sequence Spread Spectrum (DSSS) signal is a pseudo-random noise (PN) sequence. The bits in the data signal are modulo-two added with the much higher rate PN code. This effectively increases the bandwidth of the transmitted signal to that of the modulated PN sequence. At the receiver, the received sequence is multi- plied by the same PN sequence to remove the spreading.

Narrow Band Information Signal (Before Spreading)

Spread Spectrum Signal (After Spreading)

Unspread PSD Spread PSD

Figure 11-4: DSSS Spectral Occupancy Before and After Spreading

This signal design offers a number of advantages over non-spread techniques. In addition to the capability of supporting multiple simultaneous users via Code Divi- sion Multiple Access (CDMA) on the same frequency, it provides reduced Power Flux Density (PFD), reduced susceptibility to interference (including intentional jamming) and multi-path fading, and decreased probability of signal interception. The same PN code used for spreading is also used for range measurements and time transfer without the addition of a separate modulation signal for ranging.

This S-Band CDMA technique is based on agreements concluded by the Space Net- work Interoperability Panel (SNIP), an international coordinating body active in the 1980’s and 90’s comprised of NASA, ESA and JAXA representatives planning interoperability between the data relay satellites of the three space agencies – NASA’s TDRS, ESA’s Artemis and JAXA’s DRTS. Agreements were reached between the three agencies to adopt this S-Band CDMA signal design for use in the near Earth environment. Each agency was assigned a unique set of PN codes based on the SNIP design for use by their own space missions. All three sets of PN codes were implemented in the data relay satellite ground terminal of each agency to provide interoperability with the other two agency’s mission spacecraft. The Space Network In-

169 CMLP Final Report

teroperability Panel (SNIP) document Space Network Interoperable PN Codes Li- braries (451- PN Code – SNIP) defines PN codes for the interoperable space network links. The SNIP agreements became the de facto implementing standard for international CDMA interoperability and resulted in compatible international ground and flight hardware development and instances of demonstrations and mission cross- support among the space agencies’ data relay satellites.

11.5.5.2 Supporting S-band CDMA Analysis for the Lunar Scenarios

The CMLP study looked at three simultaneous user scenarios for lunar communications and navigation. The first is the LAT2-based scenario with five simultaneous R/L S-Band communications signals and a separate navigation scenario to support LSAM precision landing and lunar surface position location assuming the planned LAT2 lunar network architecture to be in place. The LAT2 Study followed the current NASA CxP assumptions for the use of the same SN signal designs used in LEO to also be used in the lunar DTE/DFE and lunar relay satellite scenarios.

The capacity of CDMA was examined for both communications and navigation multiple access cases that can occur separately in this scenario:

Likely Worst Case Scenario of Simultaneous Links Frequency Direction Element Comment Band Data Rate Number Description (kbps)

LRS #1 LRS 2-way comm/nav transmission 1 72 LSAM precision landing scenario, LSAM LRS #2 LRS 1-way nav transmission 1 50 receiving ranging signals from LRS, LCT elements

S-band LCT LCT 1-way nav transmission 2 50

LSAM (NAV Data) Total 4 222 LRS/LCT Elements to to Elements LRS/LCT

Surface Element Normal communications 4 150 Comm case Š Scenario assumes an orbiting Orion and 4 surface elements being serviced CEV In-situ CEV communications 1 192 by the LRS at S-Band LRS S-band Surface Surface Element to to Element Total 5 792 Table 11-20: LAT S-band LRS Scenarios for Multiple Access

In any multiple access communication system it is necessary to take some precaution against multiple access interference (MAI) among the various users. Normally this is in the form of “margin against MAI,” i.e. RF transmit power provided by each user terminal in excess of what would be required in the absence of interference.

Negotiation of MAI margin should be a cooperative activity, since a power increase at any transmitter can increase the interference level observed at all receivers. Given all relevant link parameters for a group of CDMA users sharing a channel, it ordinarily is possible to compute an appropriate margin value for each transmitter to use such that each link achieves its objectives in terms of desired signal-to-noise ratio at the receiver (including margin to cover unmodeled effects). When the collective user performance demand is too close to the system capacity, the margin negotiation fails

170 CMLP Final Report

and performance saturates at a multiple-access-limited asymptote at which further power increase by all users cannot improve it. When the margin allocation can be made, analysis shows that the less demanding users – those with lower data rates, shorter paths, smaller margin requests, etc. – provide greater margin (in the dB sense) than the more demanding ones, exposing an inherent “unfairness” of CDMA in trying to service a heterogeneous user community.

For the LAT-projected normal scenario involving LRS and LCT elements, even basic CDMA over a single 6 MHz channel with a 3 Mcps chip rate does not result in pro- hibitive MAI even on a single polarization, without MAI cancellation, as shown below:

Required Margin Against MAI Per User, dB for Single Polarized 6 MHz CDMA Channel Scenario LAT LRS Return Scenario LAT LSAM NAV Scenario 1st LRS User CEV Surface User 2nd LRS range LCT range w/Comm Data Rate (kbps) 192 150 72 50 50

Margin not including 4 dB for Link Margin and 3.71 3.86 0.66 0.75 0.75 Implementation Loss

Table 11-21: CDMA Capacity Analysis: LRS Navigation Scenario

The second analysis scenario uses the LAT2 traffic model, which describes 12 surface elements expected to be in place at the completion of a multi-year lunar outpost buildup. The LAT2 model lists data rates for the surface elements but does not indicate the frequency band to be used. It can be assumed that the LAT2 architecture elements, featuring WLAN, LCT and aggregation of data on Ka-Band to reduce the number of simultaneous users at S-Band would be employed in this scenario as well, if available. In this analysis scenario, however, ten simultaneous users are assumed at S-Band with the LSAM and lunar rover at Ka-Band since the lunar lander and rovers are expected to have S-/Ka-Band capability even without the Lunar Network in place (some of the other ten users could also use a WLAN base station built into the LSAM or rover and have their data aggregated for transmission at Ka-Band). Later we consider a stressing, third analysis scenario which assumes all twelve users at S-band, including the high data rate Rover and LSAM.

The maximum achievable data rate versus number of lunar platforms can be estimated analytically as shown below, taking into consideration the BER performance required for the lunar exploration platforms, the signal structure utilized for the uplink and downlink, and the implementation loss limits. This analysis result was the

171 CMLP Final Report object of a study described in a memo [“Maximum Achievable Data Rate for Lunar Exploration Mission S-band Command and Telemetry Links with an Increasing Number of Active Lunar Platforms, Revision 2”, ITT, 23 July 2007]. Interference cancellation would improve the achievable data rates shown. The assumptions of this study were as follows:

 LRO required uplink and downlink BER performance is 10-5.

 Required Eb/No for rate ½ convolutional code is 4.5 dB.

 RLEP uncoded uplink implementation loss is conservatively assumed to be 3.0 dB.

 RLEP rate ½ convolutionally coded uplink implementation loss is assumed to be 2.0 dB.

 RLEP downlink implementation loss is assumed to be 2.0 dB.

 All links must maintain a 2 dB margin against the minimum C/(N0+I0) required to achieve 10-5 BER performance for that particular link.

300 Note: • Curves generated 2 Lunar Platforms analytically assuming PN, 250 carrier, and symbol synchronizer acquisition 3 Lunar Platforms has already occurred • Curves based upon the requirement that a 2 dB 200 link margin be maintained • See Section 2 for all 4/5 Lunar Platforms other applicable assumptions 150 6 Lunar Platforms (kbps)

8 Lunar Platforms 15 Lunar Platforms 100 10 Lunar Platforms Minimum-expected downlink C/No per LRO RF ICD link 50 budgets. See Section 3.2 20 Lunar Platforms for this derivation. Maximum AchievableRate per User Data 0 40 45 50 55 60 65 70 75 80

C/N0 (dB-Hz)

Figure 11-5: Analytically-Estimated Downlink Maximum Achievable Data Rate vs. C/No Assuming Dual Polarization Used

172 CMLP Final Report

The impact caused by MAI for this second scenario is shown below on the DTE return link for the 10 S-band users on a dual-polarized, 6 MHz channel, with the high data rate CEV and LSAM at Ka-band.

DTE Return Scenario: FDMA for 2 HDR Users, Dual Polarized for Other Users with 5 Spread Users on Each Polarization: Stressing Case for 4x250 kbps, 1x10 kbps Users

Surface Extraction User EVA Suit Carrier Unit Data Rate (kbps) 250 250 10

MAI Margin (dB), not including 4 dB for Link Margin and 5.32 5.32 6.18 Implementation Loss Table 11-22: MAI Margin for DTE Return Scenario

The graph above indicates the number of S-Band R/L user platforms that can be supported simultaneously at the maximum achievable data rates indicated. The LAT2 Study nominally chose five surface platforms at 150 Kbps each, but it can be seen from the previous graph that more platforms can be supported at lower data rates. If an MAI cancellation capability is implemented in the EBGS (or LRS) multi- channel receivers, higher data rates than shown in the graph will be possible.

The NASA Constellation Program (CxP) has chosen to baseline for use at the moon as well the same SN signal designs used in Earth orbit for Cx vehicles. The SN S- Band signal designs include a fully spread (DG1 Mode 1) CDMA signal for reliable mission operations/TT&C link and a partially spread (DG1 Mode 3) CDMA signal when higher rate mission data must also be accommodated. Both these modes incorporate 2-way range and Doppler capability and time transfer service. For higher mission data rates at S-Band, unspread (DG2) is used which can only support 1-way and 2-way Doppler but neither ranging nor time transfer, both of which require the PN spreading code.

The Table below contains the CxP vehicle data rates as well as the maximum data rates that can be accommodated by the SN 3 Mcps spreading code using a 10:1 per channel chip rate to symbol rate ratio and a 6 MHz spectrum allocation. The maximum data rates with traditional Rate ½ coding as well as higher rates possible with Rate 7/8 coding or uncoded operations are shown. (It should be noted that the currently planned Cx vehicle Ka-Band maximum data rates are 6 Mbps forward and 25 Mbps return for Orion and 25 Mbps forward and 150 Mbps return for the Lunar

173 CMLP Final Report

Lander. The lander would easily provide adequate Ka-Band capacity to relay all the LAT2 surface traffic assumed in the stressing S-Band scenario to Earth using a WLAN and its Ka-band communications capability deployed as an LCT precursor.) In cases where a high data rate S-Band link is necessary, such as from a moving Rover, DG1 Mode 3 could be used to support the 1.75 Mbps return link using a Rate 7/8 FEC code simultaneously with an autonomous navigation tracking signal for IMU calibration while moving, if desired. Both the 1.558 Mbps forward link and 1.75 Mbps return link Rover data rates could be supported on S-Band using DG2 while moving, but without ranging.

SN Signal Design - Maximum S -Band Data Rates SN Modes CxP Vehicle Rates Maximum S -Band Data Rates for SN Mode (Kbps) R 1/2 R 1/2 R 1/2 R 1/2 R 7/8 R 7/8 Uncoded Uncoded Tracking Constraints/Comments F/L R/L F/L R/L F/L R/L F/L R/L

DG1 Mode 1 72 192 150 300 262.5 525 300 600 2w R & D 3 Mcps PN code rate Fully spread 3 Mcps PN code rate DG 1 Mode 3 72* 1000 150 1500 262.5 2625 300 3000 2w R & D Partially spread 6MHz spectrum allocation

DG 2 (Non -coherent) 1000 1000 3000 3000 5250 5250 6000 6000 1w D 6MHz spectrum allocation Coherent turnaround mode could provide 2w Doppler * May require 1000 Kbps on F/L requiring use of DG 2 or a differ ent DG1 Mode 3 design than used in traditional flight transponders

Notes: Earth S -Band spectrum allocation of 6MHz is assumed in the Table. Larger S -Band allocations may be possible in the lunar environment

Table 11-23: Space Network Signal Design for Maximum S-band Data Rates

Likely Worst Case Scenario of Simultaneous Frequency Data Rate Ranging Links Direction Element Number Comment Band (Mbps) Required? Data Rate Number Description (Mbps) Must assume active due to ranging LSAM 1 1.02 Yes 11.02 requirement EVA suit 4 0.01 Yes 3 0.01 3 astronauts are active Must assume active due to ranging outside of the habitat, 2 with Rover 2 1.558 Yes 11.558 requirement one of the rovers and 1 Surface Mobility Carrier 2 0.4 Yes 2 0.4 elsewhere. The 3 EVA suits O2 Excavator 1 0.2 No 1 0.2 are receiving LDR DFE O2 Mobile Servicer 1 0.2 Yes 1 0.2 command data, as are the H2/H2O Excavator 1 0.2 No 1 0.2 rover and the extraction S-band H2/H2O Extraction Unit 1 0.2 No 1 0.2 equipment. One lander is in H2/H2O Mobile Servicer 1 0.2 Yes 1 0.2 the process of landing on the Habitat 1 2.949 No Probably use Ka-band surface, receiving DFE DFE to Surface Element 0.018 Yes Likely that CEV would be handled by command data. CEV 2 0.072 Yes separate antenna beam due to spatial 1Yesseparation from other user platforms Total 12 4.408 Must be at S-band due to ranging LSAM 1 0.592 Yes 10.592 requirement EVA suit 4 0.002 Yes 3 0.002 3 astronauts are active Must be at S-band due to ranging outside of the habitat, 2 with Rover 2 1.75 Yes 11.75 requirement one of the rovers and 1 Surface Mobility Carrier 2 0.25 Yes 2 0.25 elsewhere. The 3 EVA suits O2 Excavator 1 0.25 No 1 0.25 are sending LDR DTE O2 Mobile Servicer 1 0.25 Yes 1 0.25 telemetry data, as are the H2/H2O Excavator 1 0.25 No 1 0.25 rover and the extraction S-band H2/H2O Extraction Unit 1 0.25 No 1 0.25 equipment. One lander is in H2/H2O Mobile Servicer 1 0.25 Yes 1 0.25 the process of landing on the Habitat 1 2.824 No Probably use Ka-band surface, sending DTE

DTE from Surface Element 0.024 Yes Likely that CEV would be handled by telemetry data. CEV 2 0.192 Yes separate antenna beam due to spatial 1Yesseparation from other user platforms Total 12 4.098 Table 11-24: S-band DTE/DFE Scenario for Multiple Access

174 CMLP Final Report

11.5.5.3 Stressing S-Band CDMA Analysis

The third simultaneous user analysis scenario assumes all 12 users are at S-Band, without either the LCT infrastructure or the WLAN/Ka-Band capabilities on LSAM or the rover. This scenario provides a stressing scenario for evaluation of multiple access techniques in the lunar environment and the opportunity to investigate approaches for increasing user capacity as lunar activity grows.

For this analysis the MA Team has adopted the lunar outpost communications scenario developed by LAT2, which requires simultaneous data throughput for as many as 12 users, most of whom will also require ranging. Most surface elements will require autonomous position location using the LRS. Traditional ranging from three widely separated Earth stations would only be used for IMU calibration if the LRS is not present. One MA scheme considered is CDMA over a 6-MHz channel, similar to current SN signal formats. A study was done to evaluate the ability of a traditional CDMA scheme to service a lunar user set, taking into account the effect of multiple access interference (MAI). In the study process several alternative CDMA implementations that mitigate MAI were articulated and evaluated, including a hybrid FDMA/CDMA signal design. Results of this study fed into the FOM- based analysis as the performance of CDMA was compared to that of other MA schemes for the lunar scenarios.

11.5.5.4 Scenarios/Assumptions

The analysis used newly-developed theoretical CDMA equations to calculate the transmit margin requirement for each user in an inhomogeneous CDMA user set (different data rates), as is appropriate for our lunar mission model. The traffic model, shown in Table 11-19, is that developed by the Lunar Architecture Team (LAT), derived from lunar outpost S-band traffic requirements. One scenario used to evaluate CDMA implementation of simultaneous links is the likely stressing case shown in the right part of the table. It is slightly less stressing than the full scenario shown to the left. The focus is on performance of CDMA for the DTE/DFE scenario since this is likely to be the limiting case with regard to available spectrum.

The assumptions used in the course of the analysis are as follows:

1. All surface user platforms are within the beamwidth of the LRS or Earth- based S-/Ka-Band antennas and require simultaneous communications.

2. Most platforms require ranging capability at S-band for autonomous surface position location purposes using the LRS.

175 CMLP Final Report

3. Users only differ in their required data rates.

4. LRS required uplink and downlink Eb/No is 2.2 dB with a JPL rate ½ LDPC code.

5. Per-link implementation loss is 2.0 dB.

6. All links maintain a 2 dB margin against the minimum required Eb/N0.

7. Channel bandwidth = 6 MHz, chip rate = 3 Mcps as baseline; higher rates and bandwidths considered in separate alternatives.

8. Dual orthogonal polarization signal format is the baseline.

9. PN chip transitions are asynchronous across all lunar platform uplink signals and across all downlink signals: received RF phases also randomized from user to user.

Only candidate CDMA implementations having the capacity to service the simultaneous user set despite the presence of MAI were considered in this study. The implementations were evaluated on the bandwidth they required and the amount of additional transmit margin due to MAI.

11.5.5.5 Candidate CDMA Implementations

Six implementations are considered: the first is pure CDMA, and the others are reac- tions to the inability of this first method to handle the stressing lunar multiple access scenario unless the bandwidth is much greater than 6 MHz. A summary of the results of these analyses is shown in Table 11-25.

11.5.5.5.1 Implementation #0: Pure CDMA Using Existing SN Signal Format For a single 6 MHz allocation with 12 CDMA users transmitting simultaneously at the data rates shown in Figure 11-6, the level of MAI is substantial. The channel would be driven into saturation without mitigation of the MAI levels.

11.5.5.5.2 Implementation #1: Increased Chip Rate A candidate signal format for the pure CDMA approach is shown below for the DFE link (will look similar for DTE). For a single 6 MHz allocation the channel is driven into saturation by the high data rate users (rover and LSAM), but for greater band-

176 CMLP Final Report width the links will close, even though the transmit margin requirements may be substantial.

Lander O2 Mobile Servicer EVA Suits O2 Excavator Surface Mobility 1020 kbps 400 kbps 10 kbps 200 kbps Carrier 200 kbps

Single CDMA Channel, Polarization #1 Bandwidth = 15 MHz Single CDMA Channel, Polarization #2

Rover H2/H20 Mobile EVA Suit H2/H2O Excavator H2/H20 Surface Mobility 1558 kbps Servicer 400 kbps 10 kbps 200 kbps Extractor Carrier 200 kbps 200 kbps Figure 11-6: CDMA Implementation #1: 12 CDMA Users Divided up on a Dual Polarized Single 15 MHz Channel

11.5.5.5.3 Implementation #2: CDMA/FDMA Hybrid A second CDMA implementation considered is to use a CDMA/FDMA hybrid to offload the two high data rate users onto a separate 6 MHz allocation, while employing dual polarization for the remaining ten users in order to further mitigate MAI effects. A candidate signal format for this approach is shown below for the DTE link (will look similar for DFE).

177 CMLP Final Report

Lnder 592 O2 Mobile Servicer EVA Suits O2 Excavator Surface Mobility kbps 250 kbps 2 kbps 250 kbps Carrier 250 kbps

Assigned approximately 2.3 MHz Single 6 MHz Channel, Polarization #1 Single 6 MHz Channel Single 6 MHz Channel, Polarization #2 Assigned approximately 3.7 MHz

Rover H2/H20 Mobile EVA Suit 2 H2/H2O Excavator H2/H20 Surface Mobility 1750 kbps Servicer 250 kbps kbps 250 kbps Extractor Carrier 250 kbps 250 kbps Figure 11-7: CDMA Implementation #2: CDMA/FDMA Hybrid: 2 HDR Lunar Users on Separate FDMA Channel, 10 CDMA Users Divided up on a Dually Polarized Single Channel

11.5.5.5.4 Implementation #3: Reduced Data Rate of 100 kbps for Rover and LSAM

A third implementation considered for the lunar scenario is to keep pure CDMA for all 12 users on one dual-polarized channel, but reduce the required data rates of the Rover and Lander. The premise for doing this is that these are really contingency links whose main purpose is to provide navigation capability. Consequently they may be able to get by with lower data rates.

11.5.5.5.5 Implementation #4: 10 dB Interference Cancellation Employed to Mitigate MAI

The benefit to be gained from eliminating or even reducing the multiple access interference (MAI) among a set of CDMA users has been recognized for many years. Un- like deterministically orthogonal multiple access methods, in which the user signals sharing a band can be made strictly non-interfering to within the implementation limits of achieving waveforms that are non-overlapping in space, time, or frequency, CDMA is a statistically orthogonal technique. Although spread spectrum codes assigned to different users can be strictly orthogonal at the receivers if relative time and frequency controls can be maintained—up to a hard limit on the number of such codes—only certain special circumstances permit this. In general CDMA relies on

178 CMLP Final Report

the statistical orthogonality of long, pseudorandom codes for separation of user signals, and this separation is incomplete. There is a statistically predictable amount of residual from each CDMA signal whose magnitude depends on a host of signal parameters.

Conceptually it is possible to mitigate MAI by some degree of cancellation of the interference caused to a given user by all or a selected subset of users in the CDMA channel. Algorithms that accomplish this reside at the receive terminal and involve a certain amount of infrastructure burden. In the Earth-lunar trunk line scenario (DFE/DTE), for example, the burden is placed at the Earth station in the return direction (DTE) and may thus be assumed to be relatively inconsequential. In the forward direction (DFE) the burden is assumed by each lunar user platform, with perhaps greater impact in size, weight, power, and cost.

Decreased MAI manifests itself in increased total throughput capacity for the CDMA channel at a stipulated performance level. That is, the number of users and the sum of all users’ supportable data rates increases. In lightly loaded CDMA systems, the initial MAI impact is not large, and use of elaborate means to decrease it is typically not justified. For more heavily loaded systems, particularly those operating near capacity, it becomes reasonable to conduct the cost-benefit trade to determine whether use of cancellation is appropriate.

At the top level, cancellation algorithms separate into two classes according to whether the incoming set of interfering codes is known or unknown. A receiver knowing all other user codes may employ a correlation processing channel for each to determine its approximate epoch, phase, frequency offset, etc., and follow up by subtracting a replica of the code from the received signal. But if the codes are unknown some type of estimation algorithm is required, often operating in a multi- stage mode, the depth of cancellation presumably improving at each cycle.

Cancellation methods for known codes may be further categorized as one of two fundamental types: parallel and successive. Parallel algorithms work on canceling all signals at once, whereas in successive interference cancellation, one signal, usually the largest, is estimated and subtracted, a process that is then repeated for the largest remaining signal, etc. Numerous variations on all these techniques can be found in the open literature.

Cancellation schemes vary widely in performance. Typically it is the largest interferer that is easiest to cancel, since its properties may be estimated with greater fidelity than those of weaker signals, the goal being to reduce an interfering signal to the point where its correlation peak becomes indistinguishable in the receiver noise, at which point further improvement in cancellation depth is unnecessary.

179 CMLP Final Report

The degree to which cancellation improves a CDMA system or makes CDMA the signal format of choice in a design tradeoff depends strongly on the specifics of the application. Relative to the NASA scenarios for the Moon and deep space, it is reasonable to assume that the codes are known everywhere. There exists limited simulation evidence that in the lunar proximity scenarios, even a modest amount of uniform cancellation, about 10 dB, will make many of the otherwise unworkable scenarios feasible for CDMA within a 6-MHz channel.

It has also been observed that attempts at selective cancellation of “large” signals only—done in the name of processing simplicity—tend not to be highly successful. Though it is tempting to contemplate cancelling only the highest rate users in an inhomogeneous user mix, the residual impact of the lower-rate users can remain significant. Thus non-selective cancellation seems to offer superior performance potential.

11.5.5.5.6 Implementation #5: CDMA/TDM for Forward Link

The forward link from the Earth to the moon includes the possibility of supporting multiple platforms in a point-to-multipoint fashion. We assume the transmissions can come from one antenna whose beamwidth is sufficiently wide to include all possible lunar traffic destinations.

11.5.5.5.6.1 CDMA/TDM Overview

CDMA is one natural candidate to support such service, particularly in light of the applicable bandwidth restrictions at S-band. If multiple CDMA waveforms are linearly superposed, the resulting waveform has a time-varying envelope whose peak power requirement could potentially be as large as the square of the sum of the am- plitudes of the component waveforms. A linear amplifier is required to transmit it without distortion. If each of N waveforms carry a common data rate, the peak-to- average-power ratio would be N. That is, the peak power of the linear amplifier would have to be N times the average power to insure that no waveform distortion occurs. In practice, a lower peak level can be used with some tolerance for occasional distortion. A linear HPA poses a potentially significant burden issue for both Earth stations and lunar relay satellites.

A topic that always must be addressed in CDMA is mutual interference among the waveforms, or multiple-access interference (MAI). The worst-case MAI arises when all waveforms occupy a quadrature component in common and are chip- synchronous. To minimize MAI, the transmitted chip epochs and RF phases should be randomized. This randomization happens naturally in a system where each

180 CMLP Final Report

transmitter is a separate entity, but when the waveforms are transmitted from a single source the randomization must be accomplished intentionally at the transmitter.

As an alternative, a constant-envelope form of CDMA could be used. Through nonlinear multiplexing that has its origins in the concept of selection by majority vote, multiple CDMA signals can be combined into a constant-envelope waveform in a manner that is transparent to each receiver. That is, receiver processing is not affected by whether the linear or nonlinear multiplex has been used at the transmitter.

Because of nonlinearity, constant-envelope CDMA suffers a multiplexing loss that is common (in the percentage, or dB sense) to all users. There is no small-signal sup- pression phenomenon such as will occur if a hard or soft limiter is used. In a well- designed system the loss may typically be around 2 dB, but larger and smaller values are possible. The exact loss in any given case is a function of a number of parameters.

Constant-envelope CDMA sets the peak-to-average power ratio to 0 dB and eliminates the necessity of a linear amplifier. To achieve performance on a par with the linear case the required transmit power is the average power plus a few dB to compensate for the multiplexing loss and perhaps any significant difference in MAI. This typically results in significantly lower peak power and a less expensive HPA.

The multiplexed codes need not all have the same power level. In the Earth-to-moon case where the distances are essentially equal, this means multiple data rates can be transmitted simultaneously.

Like other CDMA methods, constant-envelope CDMA incurs MAI. RF phase randomization can be used to minimize it as long as the consequent multiplexing loss is suitably small. Chip epoch randomization would call for making multiple majority votes per chip and is assumed impractical.

The multiplexed codes occupy the same bandwidth as do a linearly multiplexed set.

Codes multiplexed nonlinearly into a constant-envelope waveform may be used for ranging just as linearly multiplexed codes may. The difference in relative performance can be assessed by comparing the corresponding link budgets and properly accounting for the losses.

One point-to-multipoint CDMA scheme that makes use of existing SN signal formats is described here. This approach uses TDM and a single ranging PN code for all the users.

181 CMLP Final Report

11.5.5.5.6.2 CDMA/TDM Implementation

On the forward link use the existing Command/Range signal structure except dis- able the Command PN code. For the first CMD/RNG signal, the I channel aggre- gates the two high data rate users (Rover and LSAM) into a single unspread 2.578 Mbps TDM data stream. The Q channel of this signal transmits the common PN range signal for these two users without any data. For this first signal, tentatively use the standard 10:1 power ratio, putting more than 90% of the power on the I channel for the HDR TDM stream.

For the second CMD/RNG signal, utilize the phase coherence of the station and place the unspread, aggregate low rate data (1.83 Mb/sec) on the same channel as the first signal’s range code signal. This second signal should not need the accompa- nying range signal.

Negligible interference will occur between the range signal and the aggregate low rate data. The PN spread range signal benefits from a very low C/No requirement plus the spreading of the single HDR TDM interferer at the receiver over 6 MHz. This implementation is not listed in the Space Network User’s Guide (SNUG) as an available forward signal format, but it is a modification of SN compatible signal formats that utilize PN ranging codes. This scheme would impose potentially significant user burden, as all users would be required to receive at their relevant TDM rate.

2 users aggregated 2.578 Mbps on TDM stream 1 W

I Single PN Ranging Signal for Rover, LSAM 0.07 W

OQPSK/PM

0.1 W

+ Q

10 users 0.7 W aggregated on TDM stream 1.83 Mbps Utilizes an algebraic summation (does not spread the aggregate low rate signal).

Figure 11-8: CDMA Implementation #5: CDMA/TDM for Forward Link

182 CMLP Final Report

11.5.5.6 Results of S-band CDMA Analysis for the Lunar Scenarios

The S-band CDMA capacity analysis shows that CDMA is a very viable MA approach for the lunar operational scenarios, including the DTE/DFE case where spectrum is more tightly constrained. Although the margin computations show that pure CDMA over a dual-polarized, 6 MHz channel with a 3 Mcps chip rate cannot handle the hypothesized stressing lunar multiple access scenario used in this study – primarily due to the impact of the two high data rate users – there are several mitigation strategies that allow CDMA to service the user population adequately:

 Increase the chip rate o Potential increase up to 5 Mcps even with current SN equipment  Alleviate MAI by removing two high data rate users and servicing them in some other way (separate S-band channel or Ka-band) o Separate S-band allocation o Restrict to Ka-band for primary use except at times when ranging/navigation absolutely required o For rover, if sufficient power available, provide high data rates on Ka- band concurrently with the tracking on S-band  Consider reducing data rates of the two problematic users to values more compatible with CDMA chip rates of 3-5 Mcps  Use interference cancellation to mitigate MAI  Employ mitigation strategies not considered in this report, e.g., o Use rate 7/8 coding, which is more bandwidth efficient than rate ½ codes considered here o Orthogonal CDMA (possible only on links where all user signals originate at a single source)

Table 11-25 displays the key metrics of performance for the CDMA implementations discussed above: the margin against MAI required for each user (in excess of required 2 dB link margin), the S-band bandwidth requirement, and the major impact of using the approach.

183 CMLP Final Report

Required Margin Against MAIPer User,dB Scenario Direct-from-Earth (DFE) Direct-to-Earth (DT E) Required Surface Extraction Extraction Bandwidth Impact of CDMA Implementation User Rover Carrier Unit EVA Suit Rover Surface Carrier Unit EVA Suit per Direct ion (MHz) Data Rate (kbps) 1558 400 200 10 1750 250 250 10

6.0 = 3 Mcps Chip Rate Saturation Saturation Saturation Saturation Saturation Saturation Saturation Saturation

10.45 12.28 12.70 13.13 15.78 18.17 18.17 18.71 10.0 G reater bandwidth allocation required Change in signal format required 5 Mcps Chip Rate =

3.33 4.71 5.00 5.30 3.62 5.40 5.40 5.77 15.0 1. Higher Chip Rate on Single 7.5 Mcps Dually-Polarized 6 MHz Channel 6 MHz Dually-Polarized Chip Rate =

N/A 4.63 5.26 5.95 N/A 5.32 5.32 6.18 12.0 G reater bandwidth allocation required 6 MHz 6 MHz 2. Use Two Two 2. Use Allocations Dually-Polarized Dually-Polarized

13.62 13.31 13.94 14.64 15.68 15.84 15.84 16.70 6.0 Rover Data

Rate = 300 kbps Rate Impact on mission operational scenario

7.40 6.41 7.04 7.74 7.74 7.22 7.22 8.09 6.0 LSAM on Single Dually- Polarized 6 MHz Channel Rover Data 3. Reduce Data Rates of Rover, Rover, of Rates Data 3. Reduce Rate = 100 kbps Rate

0.48 0.91 0.98 1.06 0.49 1.04 1.04 1.14 6.0

10 dB 10 dB DT E: low (ground infrastructure implements

Cancellation cancellation) DFE: high (cancellation burden falls to lunar users)

6 MHz Channel 6 MHz 2.21 3.36 3.60 3.83 2.35 3.84 3.84 4.13 6.0 Candidate CDMA Implementations That Allow MAI Loss to be Overcome for All User Platforms 5 dB 4. Use Non-Selective Non-Selective 4. Use Interference Cancellation Cancellation Interference on Single Dually-Polarized Cancellation Notes: 1. Rover data rate of 1558 kbps/1750 kbps does not apply for the 3rd implementation 2. This table shows the required MAI margins for the worst case polarization of a dually polarized system 3. For implementation #2, the Rover and the LSAM are FDMA on another dually polarized 6 MHz allocation (one per polarization) 4. Shaded rows indicated implementations rated in MA FO M analysis Table 11-25: Summary of Results for S-band DTE/DFE CDMA Implementations

184 CMLP Final Report

11.5.6 GMSK/PN Multiple Access for Lunar Missions

This section briefly describes two different GMSK/PN modulation techniques and their motivations. Both techniques permit multiple channels by means of frequency-division multiple access (FDMA), and both techniques permit simultaneous ranging and bandwidth-efficient data on each channel. Although only one of these techniques is analyzed in this report, both should be developed and analyzed further.

11.5.6.1 GMSK/PN Overview

In the past, channel bandwidths have generally been wide enough to prevent overlap between the ranging and data signals of different users. This approach requires sufficient channel bandwidth to separate the ranging signals, which (for low data rates) can be much wider in bandwidth than the data signals. Given the limited amount of spectrum available to service a larger projected user base at potentially higher data rates, more spectrum-efficient techniques are needed.

The new GMSK/PN techniques may make it possible to support ranging with channels separated only as needed to prevent interference among user data signals. This can perhaps best be understood by considering the power needed for ranging signals and the power needed for data signals. In a receiver thermal noise-only environment, ranging signal power to noise density PR/N0 typically must be at least 0 dB-Hz for ranging measurements. In contrast, data signal power to noise density ratio PD/N0 must be at least RbEb/N0, where Rb is data rate and Eb/N0 is the required energy per bit divided by the noise power spectral density. For all but the lowest data rates, required PD/N0 is orders of magnitude greater than required PR/N0. For example, if Eb/N0 is 2 dB (typical of modern codes) and data rate is 20 kbps, required PD/N0 is 45 dB-Hz, more than four orders of magnitude greater than needed for ranging. This means that if we have the flexibility to adjust total power PT = PR + PD so that PR is much less than PD, nearly all of the power can be put into the data signal while still leaving enough power in the ranging signal to make the necessary ranging measurements.

In practice, existing integrated communications and ranging systems (other than CDMA) work just this way – normally, only a small amount of power is put into the ranging portion of the signal; most of the power is in the data signal.

GMSK, the modulation selected earlier in this report for the links for which we are considering multiple access, does not have a residual carrier, and phase-modulating a GMSK carrier with a ranging signal produces intermodulation products between ranging and telemetry but no pure ranging sideband. Peter Kinman of California State Uni- versity, Fresno, under contract to JPL, has suggested a slight modification to GMSK that produces a residual carrier and permits the addition of a pure ranging sideband

185 CMLP Final Report

(through phase modulation). The combined signal is constant envelope and the spectra of the ranging and telemetry sidebands overlap. This means, of course, that interference will be present – both to the ranging signal by the data signal and to the data signal by the ranging signal. Fortunately, as noted above, the ranging signal level can be made much lower than the data signal level, so there is very little interference to the data signal by the ranging signal. Also as noted earlier, ranging requires a very low PR/N0 to work – on the order of 0 dB-Hz. Assuming as before an Eb/N0 of 2 dB for a rate ½ code, required ES/N0 would be -1 dB. For GMSK, the maximum power spectral density of the data sideband is approximately 2PDTS = 2ES, where TS is the binary symbol rate. There- fore, the maximum power spectral density of the data sideband is approximately 2 dB greater than the noise spectral density, and the effective noise floor (including noise plus telemetry spectral density) at the center of the data spectrum is approximately 4 dB greater than that for noise alone. The ranging signal can be expected to be usable if PR can be brought above N0+2ES, i.e. if PR > (N0+2ES). This may require a slight increase in the portion of total power PT that must be put into PR, but PD is so much larger that this should have a negligible effect on PD.

This technique may allow FDMA users employing efficient modulation to operate in channels separated only as needed to prevent overlap of the data portions of their spectra; notional power spectral densities might appear as in Figure 11-9. Since data spectra will not overlap, and most of the power of each user’s transmission is in the data signal, each user’s data will encounter very little interference – a small amount from its own ranging signal and an even smaller amount from other users’ ranging signals that spill outside their allocated channels.

Figure 11-9: Notional spectra of tightly packed GMSK/PN signals with narrowband data and a low PN ranging signal floor

Contrast this to CDMA, where the data signals of all users overlap and thus all the power of every other user appears as interference, maximizing mutual interference. It may be possible to cancel some of this interference; indeed, cancellation of a single in-

186 CMLP Final Report

terferer has been demonstrated experimentally. However, interference cancellation of multiple interferers has not been demonstrated and would add considerable complexity. Polarization diversity represents a second interference mitigating possibility, pending a valid assessment of achievable polarization isolation.

Richard Orr and Dariush Divsalar have developed a related GMSK/PN technique for integrating ranging with efficient data modulation that puts the ranging signal on a subcarrier, which may be sinusoidal or a square-wave, as shown in (1).

 h  ( sin) c tGtts )(  sub)( sttPN (1)  2 

In (1), c denotes the carrier frequency (rad/s) and s the subcarrier. tG )( is the GMSK phase waveform.

For both GMSK/PN techniques, the PN chip itself can be rectangular or a half-sinusoid. The PN code may be a pseudorandom sequence or a sequence of the type devised for JPL's regenerative PN ranging transponders, i.e., a square wave with occasional chip phase reversals to resolve the square-wave ambiguities.

Two key GMSK/PN interference issues have yet to be addressed:

1. Near-Earth S-band interference immunity for PN codes that may have to operate at chip rates lower than that of TDRSS; and

2. The impact of that same environment upon the relatively narrow-band GMSK data signal content.

11.5.6.2 GMSK/PN for Forward Link

A second waveform class applicable to the forward link is the GMSK/PN class currently under study and introduced in Section 11.5.6.1. The various forms of GMSK/PN are individually constant-envelope waveforms, having the advantages previously cited in the CDMA discussion of section 11.5.5.3. For the forward link, however, it is necessary to consider how several of these might be multiplexed to provide multiple access from a single transmission point on Earth to a diversity of lunar vicinity users.

Multiple access for GMSK/PN may be accomplished by either FDMA or single-carrier TDM. For FDMA, the extent to which adjacent signals may overlap without appreciable interference has yet to be established in detail, and will differ among the alternatives, especially as regards whether the ranging signal is placed on a subcarrier. Because the waveform spectral tails consist primarily of the PN ranging signal component, the im-

187 CMLP Final Report

pact of the overlap of an adjacent signal would be somewhat like the overlap of two equal-strength, statistically orthogonal PN codes, normally not a situation giving rise to concern over MAI.

JPL has investigated more tightly packed overlap methods that interlace the spectral lines of the PN codes, claiming that the associated losses can be held to at most 1 dB.

If TDM is used, the single-carrier waveform can fill the available spectrum, and the various forward link signals, being under single-point control, will not mutually interfere. There is, of course, a slight efficiency penalty to be paid to account for overhead associated with the TDM frame structure. A single ranging code services all users in the out- bound direction, and so there are no gaps in any receiver’s access to it. For one-way ranging this is sufficient. If two-way ranging is desired there are at least these two options: (1) R/L CDMA – each user performs a regenerative turnaround and modulates data onto a user-unique code, making all signals distinguishable at the ground station; and (2) R/L FDMA – each user has a unique frequency turnaround ratio that separates users in frequency, but allows use of the single F/L code on the return as well without confusion.

An advantage of forward link MA via TDM is that the constant envelope character of the individual GMSK/PN waveform is retained. To date, however, no constant envelope form of FDMA multiplex of several GMSK/PN signals has been developed. Under linear superposition of frequency-offset individual waveforms, despite that all signals originate at a common point, the transmitter is required to support a waveform modulated in both phase and amplitude.

The occupied bandwidth of multiplexed GMSK/PN signals is of course a function of the waveform, multiplex method, and number of signals. For TDM the bandwidth is driven by the aggregate data rate of all users, and each user receiver must process at that rate, even though its average received data rate may be much lower. It is reasonable to expect that for a small number of signals GMSK would be more spectrally efficient than PN, and that as the number of signals grows the opposite becomes true. This holds for both FDMA and single-carrier TDM.

The issue of power flux density has not been examined for GMSK/PN, but it could be a key factor in the assessment. Interference issues of the types addressed in 11.5.6.1 are also subjects requiring investigation in the forward link case.

11.5.6.3 GMSK/PN Validation

A number of key questions involving carrier/subcarrier acquisition and tracking, symbol synchronization, mutual interference between the ranging and GMSK signals, and

188 CMLP Final Report multiple access signal formats and the associated losses resulting from superposition of several such waveforms remain open. Most of these issues can be addressed through simulations. We propose to first do software simulations, which should prove the vi- ability of these techniques in a few months. The software simulations, if successful, should be followed by hardware simulations and, eventually, the development of flight and ground systems.

A. Software Simulations

1. Develop multi-carrier signal transmission software model.

2. Simulate telemetry and radiometric performance using the multi-carrier signal and a single receiver. Configure the receiver to receive any one of 11 links. Per- form simulations initially under the assumption of perfect carrier and symbol synchronization.

3. Extend the software simulation to incorporate synchronization losses. Measure the amount of interference of the radiometric signal on the telemetry signal and vice versa. Characterize the interference as noise in order to compute the synchronization loop performance.

4. Simulate actual loops.

B. Hardware Simulations

1. Perform hardware simulations with (1) low data rate GMSK receiver, (2) slow sampler (less than 100Msps), (3) sufficient IF recording capability.

2. Implement ranging receiver for these tests (new hardware).

3. Perform multi-carrier and single receiver tests. Measure performance vs. parameter space.

11.5.6.4 Example GSMK/PN MA Signal Formats

This section presents two alternative MA signal designs for the GMSK/PN concept for two alternative lunar S-band MA scenarios. Both involve FDMA channelization in which the signals are placed as close as possible without mutual significant interference. Study of this aspect is ongoing.

11.5.6.4.1 Single 6 MHz Allocation

189 CMLP Final Report

Figure 11-10 shows how a single rover could be accommodated with GMSK/PN along with 10 other users in a single 6 MHz channel using dual polarization. Users communicating through directive antennas may have sufficient polarization isolation to permit such overlapping channels. If omni antennas are used, it may not be possible to obtain sufficient isolation to use dual polarization with the GMSK/PN techniques.

R R R R R E E Rover E E V V V V A A A A Freq.

0.25 0.03 0.25 0.030.4 3.86 0.4 0.03 0.25 0.03 0.25 MHz

5.78 MHz ALL Links use rate 1/2 coding

R R R R

O2MS H2E SMC O2E SMC H2EU H2MS Freq.

0.25 0.55 0.25 0.55 0.25 0.55 0.25 0.55 0.25 0.55 0.25 0.55 0.25 0.55 0.25 MHz

5.85 MHz Figure 11-10: GMSK/PN: Signal Format Using One 6 MHz Allocation

11.5.6.4.2 Two 6 MHz Allocations The second GMSK/PN signal design would easily accommodate a second rover as well as the habitat in the spectrum.

190 CMLP Final Report

Habitat Freq.

MHz 6.22

Ranging Ranging Ranging

E Rover E V V A A Freq.

0.005 3.86 0.005 MHz

Ranging Ranging Ranging

E Rover E V V A A Freq.

0.005 3.86 0.005 MHz

Ranging Ranging Ranging Ranging

O2MS H2EU SMC H2E LSAM SMC O2E H2MS Freq.

0.550.55 0.55 0.55 1.304 0.55 0.55 0.55 MHz

Figure 11-11: GMSK/PN: Signal Format Using Two 6 MHz Allocations

11.5.6.5 GMSK/PN Analysis Results

Initial GMSK/PN analyses calculated the MAI loss for FDMA using GMSK with PN subcarriers. These analyses show that even with just 6 MHz available, MAI loss is less than 1 dB. Presumably the loss will be equally small for the dual channel case and for GMSK with PN modulated on a carrier.

Rather conservative conditions to assure adequate separation of the ranging and GMSK signals have been derived, but these should be subjected to validity checking through additional analysis and simulation.

The ranging capability of the waveform as a function of its parameters has been established under the assumption of adequate isolation of the ranging signal from the baseband GMSK portion. There is, however, some uncertainty as to how the waveform should be parameterized to meet the applicable SFCG spectrum mask, and consequently the absolute ranging performance cannot be established until the mask definition uncertainty has been resolved.

191 CMLP Final Report

Figure 11-12 below shows the maximum achievable data rate for 11 or 20 GMSK/PN platforms (green arrows) assuming 6 MHz is available and single polarization, overlaid on the maximum achievable data rate for CDMA users (no interference cancellation) using dual polarization (Figure 11-5).

The two lunar platforms CDMA case is for no MAI, as each of the two users would be on a separate polarization and perfect polarization isolation is assumed. 11 GMSK/PN users could be accommodated within the same spectrum supporting 297 kbps without dual polarization, using a rate ½ code. Use of higher code rates permits even more users at this rate, though with a small increase in required C/N0 for each user.

11 GMSK/PN platforms Note : Curves generated 2 Lunar Platforms analytically assuming PN, CDMA carrier, and symbol synchronizer acquisition 3 Lunar Platforms has already occurred CDMA Curves based upon the requirement that a 2 dB link margin be maintained See Section 2 for all 4/5 Lunar Platforms CDMA other applicable assumptions 20 GMSK/PN platforms

6 Lunar Platforms CDMA 8 Lunar Platforms 15 Lunar Platforms CDMA

10 Lunar Platforms Minimum-expected downlink CDMA C/No per LRO RF ICD link budgets. See Section 3.2 20 Lunar Platforms for this derivation. CDMA

Figure 11-12: Maximum achievable GMSK/PN data rate (6 MHz, single polarization)

11.5.7 FOM Analysis of Lunar Scenarios This section describes the FOM analysis that was conducted for the candidate MA schemes for the lunar scenarios. While a final determination of the best scheme was not achieved in the time available, tentative rankings are presented in addition to the detailed FOM discussion.

11.5.7.1 FOM Weightings

The rationale for the weights assigned to the FOM elements is discussed in this section. Weights are numerical scores from 1 to five, where 1 denotes high importance, and 5 low.

192 CMLP Final Report

Capacity/Expandability: 1 This FOM is considered important in the DTE/DFE scenario because it measures the total capacity of each MA scheme which might be available to future growth in lunar missions.

Provide Radiometrics for Navigation: 1 Ranging and navigation are considered to be requirements for S-band lunar missions.

Alignment with International Standards: 3 Moderately important due to the international nature of the lunar mission, but lack of heritage is not an insurmountable obstacle.

Robustness: 2 Fairly important due to disruptive influence of multipath and interference.

Power Efficiency: 4 for forward link, 1 for return link (DFE/DTE); 1 for LRS scenarios Power requirements are not very restrictive on a forward link since they are easily controlled at the ground station, but higher transmit power from the moon and in the local lunar environment imposes a high cost.

Spectrum Utilization: 2 for DTE/DFE, probably 4-5 for LRS scenario Spectral occupancy in the 2025-2110 MHz and 2200-2290 MHz bands is very high, with multiple satellite users sharing every spectral portion of these bands. As a result of this use, gaining new multiple spectrum assignments in the S-band is difficult for lunar links operating direct to and from Earth. For links between lunar users and an LRS or an LCT, the entire spectrum in the cited bands may used if it can be shown that such use will not cause interference to other users in the near-Earth environment. It is expected that the isolation provided by path loss between the Earth environment and the Moon, combined with judicious design of the lunar relay links, will make use of a large portion, if not all, of these bands feasible.

Latency: 5 Although latency can be highly important, it appears that the actual latency added by any of the candidate MA schemes will be negligible, causing latency not to be a discriminator.

Technology Maturity: 2 Low maturity would possibly represent high risk and a large cost to develop MA technique for use in lunar environment, consequently it has a large FOM weighting.

Supports Legacy Missions: N/A For the Lunar environment, there are currently no legacy missions.

193 CMLP Final Report

User Burden: 2 This is given a large weight as it represents high cost and risk to the user community.

Infrastructure Burden: 3 Moderately important, but probably not a major discriminator between the various candidate schemes.

11.5.7.2 FOM Analysis for the Lunar Scenario – Pro’s & Con’s

Due to the uncertainty in the lunar architecture and concept of operations and the need for further study and development of the leading MA techniques, the FOM ranking was not completed. Pro’s and Con’s of the two techniques under study are listed below.

11.5.7.2.1 Capacity/Expandability

Capacity considers the aggregate capacity of simultaneous links between multiple elements, including the data rates that can be supported between individual elements and the number of simultaneous links.

It has been proposed that we calculate the total theoretical achievable data rate for each method and rank methods accordingly, independent of the actual missions present in each scenario.

The number of simultaneous S-Band reliable TT&C links is important. Due to the limited S-Band 6 MHz lunar DTE/DFE allocations, possibly one, dedicating spectrum to high data rate users, unless necessary, limits the number of users and the expandability for long term growth. High data rates will be primarily supported at Ka-Band not S- Band.

CDMA is limited by MAI which can be mitigated by cancellation techniques.

Frequency re-use to expand/manage the number of users may prove to be a bigger user/infrastructure burden for FDMA using frequency slots than CDMA using PN codes.

The GMSK/PN achievable data rate advantage in this study depends on a large frequency allocation to allow more efficient packing assuming other adjacent GMSK/PN signals. Other frequency assignments may not allow similar packing efficiencies.

Although CDMA capacity can be increased by techniques such as increasing the chip rate or using interference cancellation, capacity suffers in relation to TDMA or FDMA

194 CMLP Final Report channelizations due to the limiting nature of self-interference on the channel. The increased transmit margins required for CDMA schemes are reflected in the power efficiency FOM.

In the case of dual-polarization, CDMA may experience less reduction of capacity due to cross-polarization interference than FDMA.

11.5.7.2.2 Provide Radiometrics for Navigation

Radiometrics may be required on various links to provide traditional navigation, autonomous navigation for precision landing and position location on the moon or radio science. We will assume appropriate requirements for these dependent on the link. One approach to ranking this FOM is to determine the ranging error for each MA technique.

A TDRSS CDMA tracking accuracy study was performed [“Earth-Based Tracking of Lunar Platforms Using a TDRSS-Like CDMA Signal”, 17 October 2007] indicating that continuous ground-based TDRSS-style tracking is capable of achieving tracking measurement errors (range error and range rate error) which are generally smaller than that achieved during the Lunar Prospector mission using continuous DSN range tone tracking.

Other things being equal, range error depends inversely on chip rate. For all the methods under consideration the total transmit power is proportional to data rate and thus ranging accuracy improves inversely with the square root of data rate. A memo written by SATEL LLC [“Tracking the GMSK/PN Waveform”, 8 November 2007] described research into methods for GMSK carrier tracking, extended those to subcarrier tracking, and took a first-order look at how multipath might impact their performance. All the tracking functions seem practical, and their susceptibility to multipath is not unlike that of other approaches, e.g. TDRSS-style CDMA. It is doubtful that one would find any major discriminator between waveforms based on these issues.

On the other hand, single-channel CDMA has a near/far problem in the LSAM precision landing scenario in the face of slant range variations between the LRS-LSAM and LCT-LSAM links up to 20-30 dB. While it may be possible to adjust incoming power levels in a CDMA scheme (for instance, by using interference cancellation), this would add cost in order to implement this navigation scenario. Use of CDMA cellular- telephony-like transmit power control methods at the LCT might be feasible when the LSAM is the only user, but seems impractical with multiple users simultaneously ranging to the LCTs.

In the precision landing case, the LRS and LCT are ranging to the LSAM – users aren’t ranging to the LCT. Rover might be using an LCT 1-way range if it’s in view but that could be suspended for a landing.

195 CMLP Final Report

The Communications and Navigation Demonstration on Shuttle (CANDOS) experiment implemented an ephemeris-based (and radar-based) attenuation of a SN signal transmitted from Dryden to enable a simultaneous TDRS/GN signal reception on STS of a simulated range safety signal over two separate paths. The precise timing and ephemeris distribution planned for the LRS & LCT Lunar Network could implement a similar approach.

The CANDOS Experiment is a major plus for the autonomous nav portion of this FOM for prototyping and flying in space a multi-channel receiver providing proof-of-concept for a number of key features required for the LRS/LCT precision landing & position location scheme proposed by LAT2.

CANDOS:

 flew as a Shuttle experiment a multi-band/multi-channel transceiver - Low Pow- er Transceiver (LPT) which simultaneously received a CDMA signal from TDRS and a CDMA signal from a ground station compensating for a 20dB higher power from the GT using the power control scheme described above.

 demonstated the multiple S-Band CDMA signal reception capability needed for autonomous precision lunar landing and position location on the lunar surface.

 demonstrated onboard navigation using L-Band GPS signals simultaneously received by the LPT with the S-Band signals.

 provided the LPT engineering model that was used to perform the MA interference cancellation laboratory demo later in 2003.

The multi-channel receive capability needed for the LAT navigation scheme may also provide the interference cancellation capability needed to increase CDMA user capacity as well as solve the near/far problem during landing. (Lab demo was of one user and one interferer – capability to solve near/far problem for precision landing would be four users primarily for tracking and two interferers.)

11.5.7.2.3 Alignment with International Standards

Alignment is very difficult to measure. We decided to measure this as a subjective rank measuring how close a technique is to becoming an international standard. If, for example, all aspects are already standards, then this FOM is a 1. Non-standard waveforms should get “partial credit” for conformance to standards as proposed. The CDMA techniques have a heritage in the form of the SNIP standard, a library of PN codes agreed to

196 CMLP Final Report

by three international agencies in the past. While there is heritage in PN codes and in GMSK, the hybrid of GMSK/PN is relatively new.

A GSFC Standard for Spread Spectrum Modulation Formats for NASA Links is in the final approval process. The SNIP agreements provide a de facto international standard that has been proposed to CCSDS as a recommendation for near Earth relays.

An example of international cross support using SNIP S-Band CDMA is the ESA Auto- mated Transfer Vehicle (ATV) that will begin re-supplying ISS in 2008. Both TDRS and the ESA data relay satellite, Artemis, will support ATV missions to ISS. The SNIP- compatible transponders and S-Band CDMA proximity link equipment for communicating between ATV and ISS were built in Spain. A vendor in France provided the CDMA equipment for the European data relay satellite ground station that communicates with the ATV via the Artemis relay satellite. The vendor in Spain also provided the technology for the SNIP-compatible S-Band transponders on the Japanese HTV vehicle that will also re-supply ISS. The European Galileo Project announced last year their intention to use SNIP CDMA from a network of ground stations to provide TT&C to the constellation of Galileo spacecraft because of the robustness of its signal design.

11.5.7.2.4 Robustness

This FOM considers robustness to short term signal disruptions, whether caused by multipath or interference, unintentional or intentional. It is not clear how to quantify the robustness of each MA technique to these signal disruptions.

In the past an interference study was conducted by Ted Berman [Interference and Spec- tral Efficiency of CDMA and FDMA Signals for S-Band Lunar Missions, 1 February 2007] comparing susceptibility to unintentional interference on Earth-lunar forward and return links for CDMA versus tone ranging FDMA.

Update Ted Berman’s interference study to determine the amount of interference degradation suffered by GMSK/PN from non-Lunar missions, as well as the amount of interference degradation caused by GMSK/PN to non-Lunar missions.

CDMA provides reduced receiver susceptibility during the de-spreading process to narrowband residual carrier interferers in LEO and from non-NASA compatible lunar orbiters and surface elements.

It is conjectured that GMSK/PN may fall in between CDMA and FDMA with tone ranging in its robustness to signal disruption, but the hard evidence is scant and not completely supportive of the conjecture. As for intentional interference, the conclusion is not likely to differ from the one for unintentional interference.

197 CMLP Final Report

CDMA is known for being very robust to multipath and interference (external as opposed to self). A cursory analysis of GMSK/PN shows that its multipath susceptibility relative to ranging is comparable to that of CDMA. The multipath impact on data is not addressed.

11.5.7.2.5 Power Efficiency

This FOM measures how much power a scheme requires to meet mission requirements.

GMSK/PN outperforms CDMA; because of self-interference, CDMA links must include built in MAI margin to overcome degradation caused by multiple users on a single channel. GMSK/PN, being an FDMA scheme, requires much less MAI margin.

The peak-to-average power ratio (PAPR) prior to the power amplifier is included in this FOM. PAPR relates to the amount of linear dynamic range required of the HPA; signals with constant envelope have PAPR = 0 dB and may be transmitted by a saturating amplifier. Greater PAPR requires linearity and a higher peak power.

11.5.7.2.6 Spectrum Utilization

This FOM simply measures the amount of spectrum required by each MA scheme to service the capacity that the MA method can support. PFD levels are in principle included in this FOM as well, but are not thought to be a factor to the extent they are most relevant in the near-Earth environment.

11.5.7.2.7 Latency

Latency will come from at least three major sources: end-to-end latency, time to acquire signals, and latency from signal processing. These are somewhat independent. The sum of these could be used for this FOM. In reality, the difference in latency for the different MA schemes is negligible. Other MA schemes rating poorer in latency have already been eliminated.

11.5.7.2.8 Technology Maturity

We estimated the technology readiness level (TRL) of the critical components of the system and made a judgment as to the overall maturity as follows:

198 CMLP Final Report

Scheme TRL Level Description

Existing CDMA 9 Actual system proven through successful mission operations: long heritage of SN use with US & international manufacturing base for space & GT HW

CDMA with Increased Chip 8 Actual system proven through successful mis- Rate sion operations: actually has been used for Shut- tle (11 Mcps), though it has not been flown for our application

CDMA with Interference Can- 4-5 Component and/or breadboard validation in cellation lab or relevant environment

GMSK/PN 2 Technology concept and/or application formulated: simulation/lab results planned for the future Table 11-26: TRL Levels of MA Schemes

If the CDMA Interference cancellation demonstration equipment were available, a test through TDRSS using two LEO users could easily be performed.

However, more than just TRL levels are required to rank the MA methods. It is also important to consider the amount of development money and time would be required to bring each technique to technological maturity. The NRE costs of such development would apply to the Technology Maturity FOM. For instance, having a single mode transponder would result in reduced NRE cost and transponder weight, size, power and operational complexity.

11.5.7.2.9 Supports Legacy Missions

This is more than a statement about supporting missions that are currently flying. At any point in time, this FOM refers to supporting the deployed missions as of that date. For example, in 2025 we will still need to support missions launched in 2020. Hence, we view this FOM as a sliding window of support. The size of the window defines the metric, along with the number of legacy features that are supported.

After over 20 years, the SN still fully supports early S-Band missions using 1st generation transponders (HST & UARS). Installing new features like MAI cancellation in the SN will improve support to legacy as well as new lunar mission users.

199 CMLP Final Report

GMSK/PN is still under development whereas CDMA is being used in space and has a long history of working in the space environment

In reality, for the Lunar environment, there are currently no legacy missions, so it has been decided to remove this FOM from the ranking process.

This FOM should be included as a measure of whether the Lunar architecture MA capability supports legacy missions – this could apply to support for Orion missions to the moon with S-Band CDMA communication systems re-used from ISS phase missions.

11.5.7.2.10 User Burden

This FOM reflects the relative costs a mission user will incur to use the scheme in question. It includes costs due to mass, power, and spacecraft components. It also includes costs for real time operation and for mission planning functions. In summary, the CMLP MA sub team agreed that the following items are part of this FOM:

1. The difficulty and cost of designing a receiver for multiple simultaneous navigation signals for the LSAM precision landing scenario. For CDMA this has been demonstrated on the STS-107 CANDOS experiment, at least for a simultaneous TDRS and ground station communications link using the SN signal design. However, mitigating the near-far problem will add user burden.

2. Multi-channel receiver needed for LSAM for autonomous landing may mitigate the cost of MAI interference cancellation implementation to eliminate the near/far effect.

3. US and international manufacturer of legacy CDMA space qualified hardware.

4. The burden associated with a user having to switch his transponder from one type of signal scheme to another when transitioning from near-Earth to a lunar environment. For example, the CEV and LSAM would need to operate in this fashion if GMSK/PN were the MA scheme of choice for the lunar scenarios. But it seems likely that GMSK/PN would require only a slight power increase and very little extra weight if there is only one front end.

5. The recurring costs of producing a type of receiver for use in the lunar environment.

11.5.7.2.11 Infrastructure Burden

200 CMLP Final Report

This is analogous to User Burden, except that the Space Communications and Naviga- tion (SCAN) infrastructure incurs these costs, as opposed to the user. It is critical to measure the required cost increases on both sides in order to make programmatic decisions on investment.

Since none of the candidate schemes require any actual changes to the existing SN infrastructure, this FOM is probably not much of a discriminator. The DSN infrastructure may be affected but we don’t know at this time how all the lunar missions will be supported.

The infrastructure involved includes the SN, EBGS, LRS and LCTs.

The SN signal design is implemented and in use. There is a better understanding of what will be required to implement that signal design in the EBGS, LRS and LCT. The TDRSS S-Band Upgrade Project (TSUP) and international equipment based on the SNIP agreements give CDMA a significant advantage over GMSK/PN which is still under development with hardware and operations concepts uncertainty

There are potentially low cost SN receiver/exciter/ranging systems for EBGS by leveraging equipment from the SN TSUP follow-on production phase. Also, there are possible low cost SNIP software upgrades to incorporate SQPN into existing U.S. and international receivers.

11.5.8 Final MA Selections There was agreement in the CMLP MA team that near-Earth communications should clearly use TDRSS CDMA signal designs for MA scenarios as currently planned by the NASA CxP and supported by cross-support agreements among the international space agencies operating data relay satellite systems. For the lunar environment the picture is evolving but currently less clear. TDRSS CDMA, a proven technology with a long history of use and flight and ground equipment availability, would also function very well in the communications and tracking environment described in a recent comprehensive NASA lunar architecture study. In stressing cases in DTE/DFE scenarios where S-Band CDMA is relied upon exclusively for both low and high data rate traffic, it was shown that hardware cancellation techniques to reduce MA interference would be beneficial to increase long term user capacity and improve the power efficiency of CDMA.

GMSK/PN, although unproven, is very bandwidth efficient with respect to the data component and will support every MA scenario we considered without needing additional spectrum. The potential for interference between its narrowband data portions and other near-Earth S-band traffic remains to be evaluated, however.

201 CMLP Final Report

Due to the uncertainty of the final lunar communications and navigation architecture, the immaturity of the GMSK/PN technique, and the possible need for a CDMA MA interference cancellation capability, we recommend that both GMSK/PN and CDMA be developed further. It is also recommended that the proposed lunar MA scenarios be better defined.

Link Description Recommended MA Schemes Comments Scenario Direction Band Traditional CDMA employing unique PN S-band codes for each user Forward FDMA-type scheme using scheduled (Uplink) services through multiple single access Ku/Ka-band an-tenna beams with unique frequency slots for each user SN Traditional CDMA employing unique PN S-band codes for each user Return FDMA-type scheme using scheduled (Downlink) services through multiple single access Ku/Ka-band an-tenna beams with unique frequency slots for each user Scheduled services through multiple Forward All ground antennas which themselves (Uplink) employ multiple frequency bands GN Scheduled services through multiple Return All ground antennas which themselves (Downlink) employ multiple frequency bands Table 11-27: MA Recommendations for NER Scenarios

Link Description Recommended MA Schemes Comments Scenario Direction Band

Each of the 2 final candidates Recommended to go forward with further contains several variations Forward S-band study of CDMA and GMSK/PN; final which themselves should be (Uplink) recommendation not ready at this time further studied as well

Ka-band Traditional FDMA DTE/DFE

Recommended to go forward with further Return S-band study of CDMA and GMSK/PN; final (Downlink) recommendation not ready at this time

Ka-band Traditional FDMA LRS/LCT Recommended to go forward with further Elements to S-band study of CDMA and GMSK/PN; final LSAM (NAV recommendation not ready at this time Data) LRS Surface Recommended to go forward with further Element to S-band study of CDMA and GMSK/PN; final LRS recommendation not ready at this time

Table 11-28: MA Recommendations for Lunar Scenarios

202 CMLP Final Report

11.5.9 Mars Mars links can be broken into two distinct classes from a multiple access standpoint: DTE/DFE links and relay orbiter links. The Mars Architecture Working Group (MAWG) took data rate requirements from the Lunar Architecture Team as a starting point.

Relay links would be at S-band and Ka-band on the moon (for operational and high-rate links, respectively); similar links would be at X-band and Ka-band at Mars. At the moon these links could be either DFE/DTE or relayed through a relay orbiter; at Mars, high data rates would only be achievable on the relay links, not on DFE or DTE links, due to the much larger Earth-Mars distance. DFE/DTE links with users on the surface of Mars will be limited to much lower data rates than at the moon, mitigating spectrum problems at Mars.

The lunar architecture contemplates a fixed Lunar Communications Terminal (LCT) that would aggregate communications from many local users into a single trunk line. The MAWG contemplates the use of a similar fixed Mars Communications Terminal (MCT), but the details of the MCT depend very much on the specifics of the surface mission. If there is to be a major static outpost with a large concentration of elements, for example, a highly-capable MCT would make sense, potentially with a high- elevation tower to maximize local coverage. On the other hand, if there is to be a mobile base, a more modest MCT would be called for, integrated into the mobile vehicle.

11.5.9.1 Mars Direct-To-Earth (DTE) Multiple Access

The existing Multiple Spacecraft Per Aperture (MSPA) system at the Deep Space Net- work (DSN) can support two simultaneous downlinks at Mars through a single antenna using Frequency Division Multiple Access. This capability could be extended to support four or potentially even more spacecraft per aperture. The key issue in determining how many spacecraft will need to be supported is whether most communications are routed through a local node, such as an MCT or a Mars Relay Satellite, or whether there will be a need to support lots of links directly to users on the surface of Mars. If communications will be principally through reliable nodes, then four spacecraft per aperture may be sufficient.

In any event, even if there are a dozen simultaneous DTE links at Mars, the current FDMA scheme is clearly the preferred approach, as can be seen from the X-band and Ka-band Mars DTE FOM charts (Tables TBD-3 and TBD-4). These links are severely power-constrained, which renders CDMA impractical. GMSK/PN will be helpful at X- band to ensure that multiple missions requiring simultaneous ranging can fit within the available spectrum.

203 CMLP Final Report

11.5.9.2 Mars Direct-From Earth (DFE) Multiple Access

Multiple access is not presently available for deep space Direct-From-Earth links (uplink). The most promising uplink MA techniques are TDM, FDM, FDM/TDM, and CDMA. Multiple access on the uplink poses several difficulties: if separate carriers are needed for each user (as in FDM and CDMA), ground stations will need new linear power amplifiers (current ground transmitters are not linear). TDM and FDM/TDM would not require linear ground station amplifiers, but with TDM, the data rate on the ground transmission would have to be reduced to the maximum rate that can be received by the spacecraft with the lowest receive system performance – presumably a spacecraft with a low gain antenna.

FDM/TDM would permit the transmission of data at a high rate on one carrier to one group of spacecraft with high gain antennas and data at a lower rate on a second carrier to spacecraft with low gain antennas. Transmission of the two separate carriers requires either separate ground stations with saturated high power amplifiers or a single ground station with a linear high power amplifier. However, if the occasions when spacecraft with LGAs need MA at the same time as spacecraft with HGAs are rare, they might all be effectively accommodated with two ground stations – one primary and the second brought in when needed.

The TDM and FDM/TDM techniques have a possibly significant added user burden in that each spacecraft receiving a TDM transmission needs to be able to demodulate and decode at the aggregate data rate for the TDM stream it is receiving, which may be considerably higher than the data rate destined for the individual spacecraft.

With the high degree of uncertainty about Mars DFE scenarios, it has not been possible for us to make a final recommendation on Mars DFE Multiple Access.

11.5.9.3 Mars Relay Links

As noted earlier, Mars relay links are expected at this time to be similar to lunar relay links. We have not, as yet, performed a separate FOM analysis for Mars relay links.

11.5.10 FOM Analysis for Mars Scenarios

We have not, as yet, performed separate FOM analyses for the Mars scenarios.

204 CMLP Final Report

11.5.11 Future Work The following analyses (in progress and incomplete and those in the planning stages) are proposed as part of future work to finalize the CMLP recommendations for MA schemes in the lunar environment:

1. Requirements and Scenarios

o A notional lunar communications and navigation architecture based on the NASA HQ LAT2 Study was used in evaluating the multiple access techniques for the lunar scenario in this study. It will be very important as this work goes forward to develop a better understanding of the developing lunar network infrastructure supporting communications and navigation at the moon as well as expected concepts for operations and expansion into an environment of both cooperating and coexisting explorers from other nations. The multiple access approaches and challenges can differ significantly depending on whether one envisions sharing a surface wireless network and common space frequency assignment or building an infrastructure to support (and coexist with) users in other frequency assignments and possibly other signal designs. Further work needs to be done in understanding and planning for the expected operating environment at the moon as part of the final MA signal design selection process.

o Along the same lines, additional work should be done to develop better understanding of the network infrastructure supporting communication and navigation at Mars, and to develop concepts of operation and potential multiple access scenarios for the Mars environment.

2. Capacity/Expandability

o A BER performance evaluation should be performed which characterizes the performance improvement of MAI cancellation for the TDRSS CDMA signal structure, including the number of interferers under the lunar scenarios

o Determine effect of cross-polarization degradation on CDMA and GMSK/PN capacity for different polarization isolations.

o What is the near/far effect for CDMA? How difficult or expensive is this problem to mitigate?

3. Provide Radiometrics for Navigation

205 CMLP Final Report

o Ranging Accuracy - Determine how integration time varies by chip rate. What is the limit of acceptable integration time for each of the lunar navigation scenarios?

o Inquire whether LSAM precision landing scenario is more difficult for GMSK/PN or for CDMA.

4. Alignment with International Standards

o Detailed analysis demonstrating that GMSK/PN meets the NTIA and SFCG masks, and, where applicable, the NTIA and ITU PFD restrictions. Consider the effect of common hardware distortions such as gain flatness, phase nonlinearity and incidental AM prior to the HPA.

5. Robustness – to signal disruptions, whether caused by multipath or interference, unintentional or intentional.

o Update Ted Berman’s interference study to determine the amount of interference degradation suffered by GMSK/PN from non-Lunar missions, as well as the amount of interference degradation caused by GMSK/PN to non-Lunar missions.

o Conduct analysis of degradation for GMSK/PN from multipath based on potential lunar outpost elevation angles.

o Determine interference degradation dependence on chip rate for CDMA

o Conduct analysis of degradation for GMSK/PN from multipath based on potential lunar outpost elevation angles.

6. Power Efficiency - peak-to-average power, MAI margin to overcome degradation caused by multiple CDMA users on a single channel, Constant envelope

o Further evaluation of the multiplexing efficiency and MAI losses for various forms of constant-envelope CDMA

7. Spectrum Utilization - spectrum constraints, such as channel bandwidth (Hz) and Power Flux Density (PFD) limits (W/m2), and bandwidth efficiency (bits/sec/Hz)

8. Technology maturity (TRL) - estimate the technology readiness level (TRL) of the critical components of the system

206 CMLP Final Report

o Perform hardware and software simulations for both GMSK/PN (as described in 11.5.6.2) and CDMA with interference cancellation based on the lunar scenarios described above to validate these techniques.

o Perform NRE cost analysis of bringing GMSK/PN and Interference Cancella- tion to TRL level 9.

o BER performance evaluation should be performed for the GMSK/PN signal structure which considers common hardware distortions.

o A signal (carrier, subcarrier, symbol synch, PN) acquisition time study should be performed for the GMSK/PN signal structure and for the compact frequency plan put forward by JPL.

9. User Burden - represents cost and risk to the user community.

o Determine recurring cost of producing a GMSK/PN receiver.

10. Infrastructure Burden – represents cost and risk to the Space Communications and Navigation (SCAN) infrastructure, including lunar relay satellites and lunar communications terminals

o Performance/cost trade of multiple antennas or beams on LRS

11.6 Interaction of Link Protocols with Lower Layers

There are a number of trade-offs to be considered with respect to link protocols and the lower coding and physical communications layers for space systems. Several of these are elaborated upon in this section.

11.6.1 Link Layer ARQ One such consideration is the application of link layer automatic retransmission request (or query), also known as ARQ, to achieve reliable, error-free data transfer. This may be traded against application of forward-error correcting (FEC) channel coding or use of a higher signal-to-noise power ratio (greater power, larger antenna, etc.)

For link ARQ, frames are retransmitted until they are received correctly, so that perfect communications (e.g. zero frame error rate) is achieved. If we let q be the frame error rate, the expected number of retransmissions (excluding the original transmission) until success will be

207 CMLP Final Report

q/(1 – q)

This means that rather than transmitting, say k frames during a communications session, the ARQ system must, on average, transmit

kq/(1 – q)

additional frames.

As an example of such a trade, the following is taken from a recent paper presented at RCSGSO 2007 [Deutsch, Doyle, Wyatt and Clare, “Reducing Operations Cost through Space Networking”]. Consider a deep space link using the NASA standard codes; an inner (7, ½) convolutional code and an outer (225, 223) Reed-Solomon code. Assume further that the interleaving depth on the Reed-Solomon code is five. The frame length is therefore 10,200 bits. For a bit error rate of p, we have

q = 1 – (1 – p)10200

Without ARQ, we would have to operate this system at p = 10-10 to ensure a low enough frame error rate. This corresponds to a bit signal to noise ratio (SNR) of 2.7 dB. With ARQ, we could operate the system at a much more reasonable p = 10-6, taking a performance loss from the ARQ of only about 0.04 dB for a final bit SNR of 2.4 dB. This is shown in Figure 11-13.

Figure 11-13: Gain from link protocol automatic retransmission query ARQ

This is a net gain of 0.3 dB, or about 7%. This 7% gain could be realized in many ways, as noted above. One possibility is to schedule 7% fewer ground supports, for a corresponding operational savings.

208 CMLP Final Report

In general, ARQ can be used in this way to compensate for the lack of powerful error- correcting codes. Consider the case of 10,200 bit frames on a channel that only has the (7, ½) convolutional code. The bit SNR required p = 10-10 is now 6.8 dB while the same performance, with ARQ, can be achieved at p = 105 with a bit SNR of less than 4.5 dB. This represents an advantage of some 2.3 dB or a factor of 3.6, a much more substantial gain. The corresponding advantage for the uncoded case is ~3 dB or a factor of two.

As an additional example, consider the case of an orbiter communicating with a planetary surface asset. As the orbiter passes overhead, the received signal power will vary due not only to slant range but to multipath fading and antenna pattern variations. While these effects may be predicted to a degree, and adaptive data rates used to accommodate them to attempt to maintain a relatively constant (and low) BER, these measures can be expected to have some occasional shortcomings. Instead of incorporating an extra, costly margin to ensure no gaps in reliability occur, link ARQ may be applied to “fill in” these intermittent periods of poor link quality (coupled in addition to adaptive rate control).

11.6.2 Virtual Channels vs. Physical Channels Link layer Virtual Channels (VCs) may be traded against the use of physical channel allocations such as use of fixed FDMA. VCs can offer greater efficiency through the statistical multiplexing of traffic that is offered dynamically. That is, the gaps in bandwidth that would occur for an individual channel assigned to one source may be utilized by another through buffering mechanisms. VCs also provide ease/flexibility of bandwidth management, since reconfiguration of allocations may be administered through software changes, while physical channels may be subject to substantial constraints. Physi- cal channels may also be limited to coarser levels of bandwidth allocation resolution, including both the changing the number of channels allocated and their distribution of bandwidths.

VCs however will require some overhead, both in terms of header bits and in processing. The additional header information is necessary to distinguish the traffic sources, and buffers and associated queuing discipline for handling the traffic frames will be required.

Closely associated with the notion of VCs is the ability to differentiate prioritized traffic types. This capability allows the effective data rate to be adjusted accordingly and guarantee that a given total volume is delivered for classes identified with higher priority. This may be traded against a system in which a higher data rate is steadily provided (i.e. stronger lower layer requirements are imposed) in order to guarantee that the same level of service is always available for the critical traffic, which otherwise would be able to preempt lower priority traffic.

209 CMLP Final Report

11.6.3 Link Layer Metadata The link layer will provide a number of services for data accountability, using metadata (e.g. timestamp, sequence number, spacecraft ID, source detail) for handling of data grouped as it has been offered by the user (vs. as a bitstream). The provision of these services will require some overhead, which should be traded against the lower layers in terms of increased requirements.

11.6.4 Dynamic Access In cases of in situ lunar/planetary links between surface users and relay orbiters, constraints often arise in the power or energy capability of the user and/or the geometric availability of the link. In such scenarios, a link protocol that can dynamically achieve communications automatically may substantially improve the efficiency of communications system in terms of the energy use. Example scenarios include:

 A surface user that establishes communications only when an opportunistic event has occurred

 A surface user that requires secondary orbiter relay while primary surface-to- surface communications are temporarily obscured such as in mobile operations

Surface-orbiter communications might persist for reasonably substantial temporal periods (such as link “sessions” associated with the result of hailing operations in the CCSDS Proximity-1 protocol), or might persist for a single link frame transmission (such as may occur in commercial Very Small Aperture Terminal (VSAT) satellite protocols).

A dynamic access link protocol, with its associated burst communications needs, will require the integration of adequate physical and coding layer mechanisms to ensure efficient and reliable acquisition of each transmission initiated. The amount of benefit of the dynamic access protocol will depend on the “overhead” allocated to the preamble and synchronization processing for this purpose.

Furthermore, the dynamic access protocol benefits may be additionally enhanced in the case where the relay orbiter simultaneously services multiple surface users in its coverage footprint, i.e., in the Multiple Access case. The application of an integrated method for dynamically allocating the shared channel resource(s) results in the Demand Access Multiple Access (DAMA) method may achieve far greater efficiency than brute-force pre-assignment of dedicated resources to each surface user (via TDMA, FDMA or CDMA). However, the amount of gain achieved depends on the (unpredictable) variability of the different users’ bandwidth needs, as well as the processing complexity associated with the DAMA implementation.

210 CMLP Final Report

11.7 Interrelation of Link-Layer Protocols to Upper Layers

The scope of the link layer is a single “hop” along what may generally be a multiple- hop communications path. Several functions have been identified that have analogs in higher layers, specifically, error control, virtual channels, and store-and-forward. It is therefore possible to eliminate these functions at the link layer and instead attempt to accomplish them using higher layer protocols. Key reasons for this approach are to minimize the number of protocols requiring implementation as well as to facilitate adher- ence to standard layered protocols (specifically Internet Protocol (IP)) so as to improve interoperability. For example, rather than using link layer virtual channels to achieve traffic differentiation, one might use IP-based Differentiated Services (DiffServ) at the network layer. Another example is the possible use of transport layer Transmission Control Procedure (TCP) rather than using link layer ARQ for error control.

On the other hand, there are justifications for incorporating such functionality at the link layer. First, fundamentally different performance and operational behavior may be achieved with link layer capabilities. It is well known for example that use of TCP for error control of a path involving a wireless link will generally yield considerably poorer performance than when link error control is applied directly on the link where it is needed. Another example is link layer store-and-forward, which will allow a particular link to be followed in spite of intermittent outages, while if this function were raised to a higher layer then the intervening router would attempt to route traffic along an alternate link interface. A second basic reason for incorporating functionality at the link layer is to enable capabilities to be implemented with a compressed protocol stack, in which intermediate protocol layers that would be null are eliminated. For example, a deep space probe with a simple link to a mother ship would not need a complex protocol stack, but may nevertheless benefit highly from disruption tolerant protocol functionality.

A full resolution of link layer protocol recommendations will require further investigation in which the SCaN Network Architecture Team is engaged and the interplay of the higher layers are considered in a comprehensive fashion.

211 CMLP Final Report

12 Final Recommendations

The team’s recommendations are summarized in the following tables found within the report.

Modulations:

Category A SN and GN missions: Table 11-7 Category A Constellation missions: Table 11-8 Category B: Table 11-9 Mars In-Situ: TBD

Codes:

Category A: Table 11-15 Category B: Table 11-16 Mars In-Situ TBD

Multiple Access Schemes

Much work still needs to be done to select appropriate MA schemes for most of the MA regimes. However, the team recommends CDMA for the Near Earth Relay re- gime.

12.1 Transition considerations

The recommendations made in this study considered support of near-term and legacy missions in addition to missions that are expect to fly during the next 25 years.

New modulations, codes, multiple access schemes, and link protocols are systems entities. One cannot simply place such functionality on a single node of the communications network (e.g. a ground station or a spacecraft) and expect to reap any benefits! There needs to be a carefully devised schedule to introduce these new capabilities so that they are in place when and where they will be needed.

Although we speak of the “NASA system”, in reality this will likely be an international system comprised of nodes from many of the world’s space agencies. This means that the transition plan has to be coordinated with the international community. The schemes that are ultimately selected by this community will need to be standardized

212 CMLP Final Report

and the corresponding expertise disseminated to whoever will need to design, build, test, and operate these systems.

12.2 Schedule

One immediate concern is the planning for possible upgrades in SCaN’s communications networks; the GN, SN, and DSN. Three main factors will influence when a new modulation, coding, multiple access, or link protocol scheme will be introduced into the SCaN facilities:

1) Technology maturity of the scheme

2) Need by the customer missions

3) Availability of funds

The team as part of the FOM analysis considered the technology maturity of the various schemes.

Customer need was based on our understanding of the Agency Mission Planning Mod- el and various analyses of future mission concepts.

For the purposes of this Section, we have made no assumptions on the availability of funds: rather we assume the funds will be available when they are needed to respond to the first two items. This is an important assumption and the schedules need to be viewed in light of this.

With these assumptions, Figure 12-1 shows a notional infusion plan for modulation schemes. The 2010-2014 column exhibits the current capability. The various new CMLP- recommended modulation schemes are shown appearing when needed. The legacy schemes disappear when they are no longer needed.

It is important to note that not all nodes in these networks (e.g. DSN or GN antennas) will require every modulation scheme. Also, distinct modulation systems may be required for different frequency bands within the networks.

However, most modulation systems available today offer a wide range of formats so the number of distinct systems may be much smaller than might be indicated by the Figure.

213 CMLP Final Report

Era 2010-2014 2015-2019 2020-2024 2025-2029 2030-2034 Space Network PCM/PSK/PM PCM/PM Forward BPSK, QPSK, OQPSK OQPSK/PM Precoded GMSK PCM/PSK/PM PCM/PM BPSK, QPSK, OQPSK Return Precoded GMSK OQPSK/PM 8 PSK

Ground Network PCM/PSK/PM PCM/PM Forward BPSK, QPSK, OQPSK OQPSK/PM Precoded GMSK PCM/PSK/PM PCM/PM BPSK, QPSK, OQPSK Return Precoded GMSK OQPSK/PM 8 PSK

Deep Space Network PCM/PSK/PM Forward OQPSK/PM PCM/PSK/PM Precoded GMSK Return 8 PSK 16-QAM OQPSK/PM

Notes: 1 Not all stations in each Network may require all schemes 2 Modulators may be required for more than one spectral band Figure 12-1: Notional modulation infusion plan for SN, GN, and DSN

Figure 12-2 shows a similar infusion schedule for coding schemes. Once again, not all nodes will need all these codes. It is assumed that each coder will be capable of processing a variety of codes within a related family. Hence, for example, the LDPC coder will be able to handle all the recommended LDPC codes.

214 CMLP Final Report

Era 2010-2014 2015-2019 2020-2024 2025-2029 2030-2034 Space Network CC(7,1/2)(3) Forward (255, 223) RS(3) AR4JA & C2 LDPC CC(7,1/2)(4) Return (255, 223) RS(4) AR4JA & C2 LDPC

Ground Network CC(7,1/2)(3) Forward (255, 223) RS(3) AR4JA & C2 LDPC CC(7,1/2)(4) Return (255, 223) RS(4) AR4JA & C2 LDPC

Deep Space Network CC(7,1/2) Forward AR4JA & C2 LDPC Turbo CC(7,1/2) RS Return AR4JA & C2 LDPC Turbo

Notes: 1 Not all stations in each Network may require all schemes 2 Coders (encoders and decoders) are assumed to be capable of coding family of code parameters 3 Currently supported by NASA SN and GN as pass-through data (i.e., customer performs encoding) 4 Currently supported by NASA SN and GN equipment (R-S decoding not supported for all SN and GN service modes) Figure 12-2: Notional coding infusion plan for SN, GN, and DSN

MA: TBD

215 CMLP Final Report

13 Conclusions

The NASA CMLP study team has examined all the links in the NASA Space Communi- cations and Navigation Architecture. For each link, the team has recommended a small set of modulation, coding, and (as appropriate) multiple access schemes. In addition, the team has recommended attributes that should be included in the selection of link protocols to use with these links (on a single-hop basis.)

The team had a great deal of support from NASA Headquarters and from all the appropriate NASA centers. The technical work was reviewed at important junctures during the study and, in fact, has resulted in a number of soon-to-be-submitted technical papers describing new technical results.

The study, in addition to providing valuable guidance to the NASA SCaN office and to the international space communications community, has been a great opportunity to share ideas between NASA centers.

The co-leaders of the study thank all the participants and their organizations for their hard work and results.

216 CMLP Final Report

14 Future Work

The choice of a recommended multiple access scheme for the lunar scenarios requires additional work, as outlined in Section 1.1. We recommend that NASA approach this by doing three things:

1. NASA continue to hone lunar scenarios,

2. NASA invests in a technology development effort for the GMSK/PN scheme, and

3. NASA invests in a technology development effort for CDMA “crowded signal scenarios”.

Additional information on this may be found in Section 11.5.11.

217 CMLP Final Report

15 Acronyms

AMPM Agency Mission Planning Model AOS Advanced Orbiting System APL Applied Physics Laboratory ARA4J Accumulate Repeat 4 Jagged Accumulate ARQ Automatic Retransmission Request AWGN Additive White Gaussian Noise BER Bit-Error-Rate BPSK Binary Phase Shift Keying Cat. A Category A (Near Earth: less than 2 million km from Earth) Cat. B Category B (Deep Space: 2 million km or more from Earth) CCSDS Consultative Committee for Space Data Systems (http://www.ccsds.org/) CD Collision Detection CEV Crew Exploration Vehicle CDMA Code Division Multiple Access CMLP Coding, Modulation and Link Protocol CRC Cyclic Redundancy Check CSI Cross-channel Interference CTN Communications and Tracking Network CTS Clear to Send CxP Constellation Program DAMA Demand Assigned Multiple Access dB Decibel DFE Direct From Earth DG1 Data Group 1 DG2 Data Group 2 DLL Delay-Lock Loop DL-SDU

Data Link Layer SDU DPSK Differential PSK DRS Data Relay Satellite DSMS Deep Space Mission System

218 CMLP Final Report

DSN Deep Space Network DSSS Direct-Sequence Spread Spectrum DTE Direct To Earth EBGS Extended Bright Galaxy Sample EDL Entry, descent and landing EIRP Equivalent Isotropic Radiated Power EMI Electromagnetic interference ESA European Space Agency ESSP Earth System Science Pathfinder EVA Extravehicular Activity FCC Federal Communication Commision FDMA Frequency Division Multiple Access FDX Full Duplex FEC Forward Error Correction FHSS Frequency Hopped Spread Spectrum FOM Figure Of Merit FSK Frequency Shift Keying GBN Go-Back-N GLAST Gamma-ray Large Area Space Telescope GMSK Gaussian Minimum-Shift Keying GN Ground Network GPM Global Precipitation Mission GPS Global Positioning System GRC Glenn Research Center GRGT Guam Remote Ground Terminal GSFC Goddard Space Flight Center GT Ground Terminal HARQ Hybrid ARQ HDR High Data Rate HDX Half-Duplex HST Hubble Space Telescope I/O Input/Output IEEE Institute of Electrical and Electronics Engineers IOAG Interagency Operations Advisory Group ISI Inter-symbol Interference ISS International Space Station ITU International Telecommunications Union JDEM Joint Dark Energy Mission JHU Johns Hopkins University JPL Jet Propulsion Laboratory JSC Johnson Space Center JWST James Webb Space Telescope LAT Lunar Architecture Team

219 CMLP Final Report

LCNS Lunar Communications and Navigation System LCT Lunar Communications Terminal LDPC Low Density Parity Check LDPCC LDPC Code LEO Low Earth Orbit LEO-P Polar Low Earth orbit LLC Logical Link Control LRS Lunar Relay Satellite LSAM Lunar Surface Access Module L-SAP Layer Service Access Point LTP Licklider Transmission Protocol MA Multiple Access MAC Medium Access Control MACA Multiple Access with Collision Avoidance MAI Multiple Access Interference MARSats Mars Areostationary Relay Satellites Mbps Mega bits per second Mcps Mega chips per second MCT Mars Communications Terminal MDS Maximum-Distance-Separable MHz Mega Hertz MIDEX Medium class Explorer MLD Maximum Likelihood Decoding N/A Not Applicable NASA National Aeronautics and Space Administration NAT Network Architecture Team NER Near Earth Relay NRE Non-Recurring Expense NRZ Non-Return to Zero NSA National Security Agency NTIA National Telecommunication and Information Administration OFDM Orthogonal Frequency Division Multiplexing OOB Out-Of-Band OQPSK Offset Quadrature Phase-Shift Keying OWLT One Way Light Time PAPR Peak-to-Average Power Ratio PCM Pulse-Code Modulation PFD Power Flux Density PFF Precision Formation Flying PM Phase Modulation PN Pseudo-random Noise PPM Pulse Position Modulation PRMA Packet Reservation Multiple Access

220 CMLP Final Report

PSD Power Spectral Density PSK Phase Shift Key QAM Quadrature Amplitude Modulation QPSK Quadrature Phase-Shift Keying QoS Quality of Service RF Radio Frequency RFI Radio Frequency Interference RLC Radio Link Control RMS Root Mean Squared RS Reed-Solomon RTLT Round-Trip Light Time RTS Request to Send Rx Receive SAR Segmentation and Reassembly SCA Space Communications Architecture SCaN Space Communication and Navigation Office SCAWG Space Communications Architecture Working Group SDMA Space Division Multiple Access SDU Service Data Unit SEL1 Sun-Earth Lagrange Point #1 SEL2 Sun-Earth Lagrange Point #2 SFCG Space Frequency Coordination Group SN Space Network SNAP SuperNova/Acceleration Probe SNIP Space Network Interoperability Panel SNIP Space Network Interoperability Panel SNR Signal-to-Noise Ratio SNUG Space Network User’s Guide SO Space Operations SOMD Space Operations Mission Directorate SOQPSK Shaped Offset Quadrature Phase Shift Keying SQPSK Staggered Quadrature Phase-Shift Keying SR Selective Repeat SRS Space Research Services SYSOP System Operator SPWG System Planning Working Group STGT Secondary TDRS Ground Terminal STS Space Shuttle Transportation System TC Telecommand CCSDS 232.0-B-1 TCDMA Time Division CDMA TCP Transmission Control Protocol TDFH Time Division Frequency Hopping TDM Time Division Multiplexing

221 CMLP Final Report

TDMA Time Division Multiple Access TDRS Tracking and Data Relay Satellite TDRSS Tracking and Data Relay Satellite System (also know as Space Network) TH-PPM Time Hopping-Pulse Position Modulation TKUP-A TDRSS K-band Upgrade Project Augmentation TM Telemetry Space Data Link Protocol CCSDS 132.0-B-1 TPC Turbo Product Codes TPF Terrestrial Planet Finder TRL Technology Readiness Level TSUP TDRSS S-band Upgrade Project TT&C Telemetry, Tracking and Control Tx Transmit UARS Upper Atmosphere Research Satellite UDP User Datagram Protocol US United States WDMA Wavelength Division Multiple Access WISE Wide-field Infrared Survey Telescope WLAN Wireless local area network WSGT White Sands Ground Terminal VC Virtual Channels VSAT Very Small Aperture Terminal

222 CMLP Final Report

16 Appendices

Appendix 1: Performance Estimates for Two-Dimensional Coded Modu- lations

Appendix 2: GN and SN Modulation Downselect Workbook

Appendix 3: Near-Earth Constellation Modulation Downselect Work- book

Appendix 4: Category B Modulation Downselect Workbook

Appendix 5: Coding FOM Analysis Workbook

Appendix 6: Comparison of C2 and TPC

Appendix 7: Ranging with GMSK

223 CMLP Final Report

17 References

i NASA Space Communication and Navigation Architecture Recommendations for 2005-2030, 15 May 2006. ii G. Noreen, P. Kinman & R. Bokulic, “Detection of Very Weak Transmissions from Deep Space,” Acta Astronautica, Vol. 39, No. 1-4, pp. 81-90, 1996. iii DeBoy, C. C., et al., “The New Horizons Mission to Pluto: Advances in Telecom- munications System Design,” 55th International Astronautical Congress 2004, paper IAC-04-M.5.05, Vancouver, Canada, 2004. iv Skolnik, M., Introduction to Radar Systems, McGraw-Hill, New York, 1980. v DSMS Telecommunications Link Design Handbook, TMOD No. 810-005, Rev. E, Module 214, Change 1, Regenerative Ranging, Jet Propulsion Laboratory, California Institute of Technology, 31 March 2004.

224