Clock Synchronization and Dominating Set Construction in Ad Hoc Wireless Networks
DISSERTATION
Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the
Graduate School of The Ohio State University
By
Dong Zhou, M.S.
*****
The Ohio State University
2005
Dissertation Committee: Approved by
Professor Ten-Hwang Lai, Adviser Professor Anish Arora Adviser Professor Dong Xuan Graduate Program in Computer Science and Engineering c Copyright by
Dong Zhou
2005 ABSTRACT
Ad Hoc wireless networks have been gaining importance in the communication world for the past decade. Wireless network extends the access to the network by removing the restriction of physical wires. Ad hoc network further improves the network coverage and availability to places without infrastructure support. Clock synchronization and dominating set are two of the fundamental issues in the wireless ad hoc networks. They are important for the correctness and/or performance of many protocols and applications. We focus on IEEE 802.11 wireless ad hoc networks in this dissertation due to the wide deployments of 802.11 networks. The theories and practices are definitely extensible to other types of ad hoc networks.
IEEE 802.11 wireless network depends heavily on the distribution of timing in- formation to all the stations in the network. Clock synchronization is important for frequency hopping, power saving and wireless medium reservation. We review the
Timing Synchronization Function (TSF) of Ad Hoc mode defined in the 802.11 stan- dard. It is well-known that the 802.11 TSF is not scalable. We carefully analyze the root causes of the scalability problem and design new schemes to overcome the problem. Our new schemes show great improvement over the 802.11 TSF and other solutions in the fields. Our solutions have nice characteristics: scalable, accurate, bounded and adaptive to station mobility. We are able to control the maximum clock differences under 25 µs and 50 µs in single hop and multihop networks respectively.
ii The performance improvement is at least 200% or more compared with the current solutions.
Dominating set has been widely used in multihop ad hoc networks (MANET) by numerous routing, broadcast and collision avoidance protocols. The problem to construct a minimum sized dominating set is known to be NP-hard. We propose a protocol that is simple, distributed, inexpensive, and adaptive to station mobility.
We show that our protocol can construct dominating set using 35% to 60% less nodes than other distributed dominating set protocols.
We study and address two of the fundamental and difficult problems in ad hoc wireless networks. Our solutions show significant improvements over the current so- lutions. Our work provides solid foundations to other important problems: routing, power saving, frequency hopping, media access scheduling, and broadcast storm mit- igation.
iii To Xiang, Ellen and Austin
iv ACKNOWLEDGMENTS
I would like to thank my advisor, Professor Ten-Hwang Lai, for his dedication, commitment, insight and advices during my research. The conversation with Pro- fessor Lai is always interesting and stimulating. His research methodology and the way he approaches the problems teach me how to meet the challenges posed by the problems and how to push my own limitations in order to land on a higher ground.
These advices will follow me through my personal and professional life.
I would like to thank Professor Anish Arora for sharing information on his sensor network projects with me. These discussions expanded my scope and gave me insight into my own research. I am amazed by the elegant theories and ideas that Professor
Arora and his team show in these projects. It propels me to set a higher standard for my own research.
I would like to thank Dr. Dong Xuan. He spent many hours with me on research problems and research methodologies. He advises me on many aspects of my profes- sional life. He shares his academic experiences and suggests me how to pick the best career path. These are valuable lessons for my next career move.
I am deeply indebted to my wife, Xiang Shen. She encouraged me to come back to finish my degree and has been supporting me all these years. Her love surrounds me through the most difficult times. Her encouragement and patience makes this dissertation possible.
v I would like to thank my two lovely children: my daughter Ellen and son Austin.
They are my inspiration for my research. They always show their love when I come home even after I have spent many sleepless days away from them.
I would like to thank my parents who always encourage me not to give up. Their persistence makes my comeback possible. I like to thank my sister to take care my parent when I am away from home. I would also thank my in-laws. They shared their research experience and showed me how to deal with stress and difficult situation.
I would like to thank Dr. Min-Te Sun, we worked together on several research papers and I admire his dedications to his friends. We become personal friends through research collaboration.
Finally I want to thank Li-Fei Huang for many of the discussions on clock syn- chronization.
vi VITA
September 11, 1968 ...... Born - Zibo, Shandong, China
July, 1989 ...... B.S. Computer Science, Beijing Univer- sity, Beijing, China July, 1992 ...... M.S. Computer Science, Beijing Uni- versity, Beijing, China December, 1993 ...... M.S. Computer and Information Sci- ence, The Ohio State University, Columbus, Ohio, USA January 1996-July, 1999 ...... Technical Consultant, Lucent Tech- nologies, Columbus, Ohio, USA July 1999-July, 2000 ...... Technical Consultant, MCI/UUNET, Columbus, Ohio, USA July 2000-present ...... Member of Technical Staff, Lucent Technologies, Columbus, Ohio, USA.
PUBLICATIONS
Research Publications
D. Zhou, T. H. Lai, A Scalable and Adaptive Clock Synchronization Protocol in IEEE 802.11-Based Multihop Ad Hoc Networks, Proceedings of the second IEEE international conference on Mobile Ad-hoc and Sensor Systems, Nov. 2005.
B. Kim, J. Yang, D. Zhou, M. T. Sun, Energy-Aware Connected Dominating Set Construction in Mobile Ad Hoc Networks, Proceedings of 14th IEEE International Conference on Computer Communications and Networks, pages 229-234, Oct. 2005.
vii D. Zhou, T. H. Lai, A Compatible and Scalable Clock Synchronization Protocol in IEEE 802.11 Ad Hoc Networks, Proceedings of the 34th International Conference on Parallel Processing, pages 295-302, Jun. 2005.
D. Zhou, M. T. Sun, T.H. Lai, A Timer-based Protocol for Connected Dominating Set Construction in IEEE 802.11 Multihop Mobile Ad Hoc Networks, Proceedings of the 2005 IEEE International Symposium on Application and the Internet, pages 2-8, Jan. 2005.
D. Zhou, T. H. Lai, Analysis and Implementation of Scalable Clock Synchronization Protocols in IEEE 802.11 Ad Hoc Networks, Proceedings of the first IEEE interna- tional conference on Mobile Ad-hoc and Sensor Systems, pages 255-263, Oct. 2004.
T. H. Lai, D. Zhou, Efficient and scalable IEEE 802.11 ad-hoc-mode timing synchro- nization function, 17th International Conference on Advanced Information Network- ing and Applications, pages 318-323, Mar. 2003.
D. Zhou, T. H. Lai, Backup Group Multiplexing: An Integrated Restoration Strat- egy for IP over WDM Optical Networks, Proceedings of International Conference on Computer Network and Mobile Computing, pages 235-240, Oct. 2001.
D. Zhou, T. H. Lai, Efficient Resource Allocation in Self-Healing Mesh Multiprotocol Label Switching Networks, Proceedings of IEEE GlobeCom, pages 2671-2675, Nov. 2001.
viii FIELDS OF STUDY
Major Field: Computer Science and Engineering
Studies in: Networking Prof. Ten-Hwang Lai Software Systems Prof. Anish Arora Theory and Algorithms Prof. Rephael Wenger
ix TABLE OF CONTENTS
Page
Abstract ...... ii
Dedication ...... iv
Acknowledgments ...... v
Vita ...... vii
List of Tables ...... xiii
List of Figures ...... xiv
Chapters:
1. Introduction ...... 1
1.1 Introduction to Wireless Networks ...... 1 1.2 Overview of 802.11 standards ...... 3 1.3 Overview of other wireless technologies ...... 5 1.4 Motivation ...... 6 1.4.1 Clock Synchronization ...... 6 1.4.2 Dominating Set Construction ...... 7 1.5 Contributions ...... 8 1.6 Organization of this Dissertation ...... 11
2. Background ...... 14
2.1 History of clock synchronization ...... 14 2.1.1 Full Plesiochrony ...... 15 2.1.2 Hierarchical Master-Slave Synchronization ...... 16 2.1.3 Mutual Synchronization ...... 16
x 2.2 IEEE 802.11 TSF and Frame Format ...... 17 2.3 Mathematical Model for Clocks ...... 20 2.4 Clock Technologies and Clock Performance Hierarchy ...... 22
3. Issues and Challenges in Wireless Ad Hoc Networks ...... 24
3.1 The importance of Timing Synchronization in 802.11 Ad Hoc Networks 24 3.2 Issues in the Current 802.11 TSF ...... 26 3.3 Requirements of for Timing Synchronization in 802.11 Networks . . 27 3.4 Formal Definition of the Minimum Dominating Set Problem . . . . 29 3.5 The Significance of the Minimum Dominating Set Problem . . . . . 29 3.6 Challenges of Solving the Minimum Dominating Set Problem in Wireless Ad Hoc Networks ...... 31
4. Analysis on the Scalability of 802.11 TSF ...... 33
4.1 Analysis of Beacon Contention ...... 34 4.2 Metrics for asynchronism ...... 36 4.3 Root Causes of Asynchronism and Some Possible Remedies . . . . 38 4.4 Summary ...... 41
5. Timing Synchronization Protocols in a single-hop 802.11 IBSS ...... 43
5.1 Overview of Clock Synchronization Protocols ...... 43 5.2 ATSP: Adaptive Timing Synchronization Procedure ...... 47 5.3 ABTSF: Adaptive bi-directional TSF ...... 49 5.4 TATSF: Tiered Adaptive Timing Synchronization Function . . . . 52 5.4.1 Overview of TATSF ...... 52 5.4.2 Analysis of TATSF ...... 54 5.5 SATSF: Self Adjusting Timing Synchronization Function . . . . . 57 5.5.1 Overview of Protocol ...... 57 5.5.2 Detailed Description of Protocol ...... 60 5.5.3 Analysis of Protocol ...... 66 5.5.4 Comparison between ASP and SATSF ...... 71 5.6 Performance Studies ...... 73 5.6.1 Simulation results for 802.11 TSF/ Bi-directional TSF . . . 74 5.6.2 Simulation results for ATSP ...... 76 5.6.3 Simulation results for TATSF/ ABTSF ...... 78 5.6.4 Simulation results for ASP and SATSF ...... 80 5.7 Summary ...... 87
xi 6. Clock Synchronization in 802.11-based Multihop Ad Hoc Networks . . . 89
6.1 Overview of Current Clock Synchronization Protocols in 802.11- based Multhop Ad Hoc Networks ...... 89 6.2 MATSF ...... 90 6.2.1 Overview of our Protocol ...... 91 6.2.2 A Scalable Solution for Clock Synchronization: MATSF . . 98 6.2.3 Analysis of the protocol ...... 103 6.3 Performance studies ...... 110 6.4 Summary ...... 114
7. A Timer-based Protocol for Connected Dominating Set Construction in IEEE 802.11 Multihop Mobile Ad Hoc Networks ...... 115
7.1 Current CDS solutions ...... 115 7.2 Requirements for Connected Dominating Set in MANETS . . . . . 117 7.3 MAC-layer Timer-based Connected Dominating Set Construction Protocol ...... 118 7.3.1 Notation Definitions ...... 118 7.3.2 The MTCDS Protocol ...... 119 7.3.3 Beacon Frame Extension ...... 123 7.3.4 Correctness of the MTCDS Protocol ...... 123 7.3.5 Station Mobility Handling ...... 126 7.4 Implementation Considerations ...... 128 7.5 Performance Studies ...... 130 7.6 Framework Extension ...... 132 7.7 Summary ...... 134
8. Contributions and Future Work ...... 136
8.1 Research Contributions ...... 136 8.2 Future Work ...... 138
Appendices:
A. ABBREVIATIONS ...... 141
Bibliography ...... 145
xii LIST OF TABLES
Table Page
2.1 Beacon generation window and slot time ...... 18
5.1 Token bits explanation ...... 51
5.2 Maximum clock offset of bi-directional TSF for n = 200 ...... 74
5.3 Maximum clock offset of TATSF and ABTSF ...... 80
5.4 Maximum clock offsets of SATSF and ASP ...... 84
5.5 Average maximum clock offset of SATSF and ASP in mixed mode . . 85
6.1 Maximum clock offset of MATSF, ASP and 802.11 TSF ...... 113
6.2 Average maximum clock offset of MATSF, ASP and 802.11 TSF . . . 114
xiii LIST OF FIGURES
Figure Page
1.1 The IEEE 802 family and relationship to OSI model ...... 4
2.1 Full plesiochrony (Anarchy) ...... 15
2.2 Hierarchical Master-Slave Synchronization ...... 16
2.3 Mutual Synchronization ...... 17
2.4 Beacon generation window ...... 20
4.1 p(n, w)...... 35
4.2 p0(n, w) ...... 36
4.3 H0 and L0 ...... 38
5.1 ATSP: Adaptive Timing Synchronization Procedure ...... 48
5.2 Beacon Frame Format for ABTSF ...... 50
5.3 TATSF: Tiered Adaptive Timing Synchronization Procedure . . . . . 53
5.4 Maximum clock offset for 802.11 TSP ...... 75
5.5 Maximum clock offset for 802.11 TSP ...... 75
5.6 Maximum clock offset for ATSP ...... 77
5.7 Maximum clock offset for ATSP ...... 77
xiv 5.8 Maximum clock offset for TATSF ...... 78
5.9 Maximum clock offset for TATSF ...... 79
5.10 Maximum clock offset for ASP ...... 81
5.11 Maximum clock offset for ASP with one non-compliance station . . . 82
5.12 Maximum clock offset for SATSF ...... 83
5.13 Maximum clock offset for SATSF with one non-compliance station . . 83
5.14 Maximum clock offset for SATSF ...... 84
6.1 Covered, bridge and uncovered nodes ...... 95
6.2 One bridge is enough to enter DS ...... 97
6.3 It is difficult to recognize gateways...... 97
6.4 Probability that success black beacons ≥ 1 ...... 108
6.5 Maximum clock offset for 802.11 TSF ...... 110
6.6 Maximum clock offset for ASP ...... 111
6.7 Maximum clock offset for MATSF ...... 112
7.1 Initiator Election Protocol ...... 120
7.2 Timer-based Connected Dominating Set Construction Protocol . . . . 122
7.3 Beacon Frame Format ...... 124
7.4 initMax for m × m grid under various beacon success rate ...... 129
7.5 DS size in 4 x 4 grid ...... 131
7.6 DS size in 8 x 8 grid ...... 132
xv CHAPTER 1
INTRODUCTION
In this chapter, we will overview the current development in wireless networks and present the motivations of our research works and highlight the contributions we have made through our research efforts.
1.1 Introduction to Wireless Networks
Wireless networks have experienced explosive growth during the last decade. The revenues from wireless networks are exceeding that of wired networks. This phe- nomenon is due to the advantages offered by wireless networks. People move, but the data is typically stored centrally. The capability to access data on the move leads to large productivity gains [16]. First generation wireless networks are analog phone networks such as AMPS (Advance Mobile Phone Services) developed by Bell Lab- oratories. Frequency division multiple access (FDMA) technology is used for radio communications. These networks supplement the wired PSTN (Public Services Tele- phone Networks). Analog wireless systems allow people to make phone calls where the
PSTN services are not available. Second generation digital networks are developed to cope with the higher user densities. GSM (Global System for Mobile Communi- cations), North American TDMA (Time division multiple access) IS-136 and CDMA
1 (Code Division Multiple Access) IS-95 are among the popular second-generation sys- tems. To meet the demand of wireless data services such as SMS (Short Messaging
Service), GPRS (General Packet Radio Service) and EDGE (Enhanced Data-rate for
GSM Evolution) are developed for TDMA/GSM systems and CDMA2000-1X system is developed for CDMA systems. To support wireless Internet access and wireless multimedia services (including audio, video, and images), third generation (3G) sys- tems are developed. Wideband CDMA (W-CDMA), cdma2000 (including EVDO and
EVDV), TD-SCDMA (Time Division Synchronous Code Division Multiple Access) are among the approved 3G standards. 3G technologies support 144 Kbps bandwidth with high-speed movement (e.g., in vehicles), 384 Kbps with pedestrian, and 2 Mbps for stationary users [26].
In this dissertation, we focus on the wireless local area networks (LAN), in par- ticular IEEE 802.11 wireless LANs. Since there is no need to install cables, wireless
LANs are extremely popular in old office buildings, historic sites and homes. Even for modem facilities, wiring the build is still expensive and time-consuming. With- out the restrictions of physical cables, mobile users can form a network without any infrastructure support. This flexibility is important for military networks and group gatherings where people are from many different organizations. This is the ad hoc mode of operation. An ad hoc wireless network is a LAN that the network devices are part of the network only for the duration of a communication session and the network is formed when these devices are in some close proximity to the rest of the network. In Latin, ad hoc literally means “for this,” further meaning “for this purpose only,” and thus usually temporary. The 802.11 networks are heavily prompted by the
Wi-Fi (Wireless Fidelity) Alliance[36]. It oversees the certification tests for product
2 interoperability. The Wi-Fi marketing effort is widely successful. Many airports, hotels, fast-food facilities, agencies, schools, homes and various other businesses have adopted 802.11 wireless LANs as alternatives to wired LANs.
1.2 Overview of 802.11 standards
IEEE 802.11 is a member of the IEEE 802 family, which is a set of specifications for
LAN technologies. The 802 standards define the physical and data link layer protocols in the OSI model. The data link layer is further divided into two sublayers: Logical link control (LLC) sublayer and media access control (MAC) sublayer. Fig. 1.1 shows the IEEE 802 family protocol stack and its relation to the OSI model[16].
The base 802.11 specification[1] supports two physical layers: FHSS and DSSS in the 2.4 GHz frequency band. The data rate is 1 Mbps or 2 Mbps. The 802.11b specification[3] adopts a high-rate DSSS physical layer in the 2.4 GHz frequency band. The data rate can be 5.5 Mbps or 11 Mbps. The 802.11a standard[2] specifies an orthogonal frequency division multiplexing (OFDM) physical layer in the 5 GHz frequency band. The data rate is up to 54 Mbps. The 802.11g specification[4] can support data rate up to 54 Mbps and operate in the 2.4 GHz frequency band. It can use FHSS, DSSS or OFDM. It is compatible with 802.11b. An upcoming protocol in the 802.11 protocol family is called 802.11n. It is not yet standardized. It will use multiple-input, multiple-output (MIMO) antenna technology. The data rate can be over 100 Mbps. The radio range of 802.11 devices is in the order of 100 meters.
In 802.11 networks, a group of stations within the communication range of each other can form a basic service set (BSS). There are two types of BSS: infrastructure
3 Data link 802 802.1 layer Overview Management 802.2 Logical link control (LLC) and LLC Architecture sublayer
MAC 802.11 MAC 802.3 802.5 sublayer MAC MAC
802.11b Physical 802.11 802.11 802.11a HR/ layer 802.3 802.5 FHSS DSSS OFDM DSSS PHY PHY PHY PHY PHY PHY
Figure 1.1: The IEEE 802 family and relationship to OSI model
BSS and independent BSS (IBSS). For infrastructure BSS, an access point is the cen- tral point of the BSS. All communications including communications between stations within the BSS is through the access point. For IBSS, the stations can communicate directly with each other and all the stations are within the direct communication range. An IBSS is an ad hoc network. In this dissertation, we will mainly study the issues in ad hoc networks.
Compared with the cellular wireless networks, 802.11 wireless LANs have higher bandwidth, lower cost but limited coverage. Cellular networks enjoy large coverage area but lower data rate. These two types of networks complement each other. The integration of these networks is one of the future research directions.
4 1.3 Overview of other wireless technologies
Bluetooth is a WLAN standard for radio devices within the range of 10 meters[42].
Bluetooth can support up to 7 devices per piconet. Frequency hopping and TDD
(Time Division Duplex) are used for point-to-point communications. The data rate is around 1 Mbps. Blue tooth device is very cheap ($5 or less). The intent is to connect devices with a nominal range of about one to three meters. It can be used to connect laptop, PDA, cellular phone, keyboard, mouse, printer and other devices[32].
IEEE 802.16 (Wi-Max) technology utilizes microwave for high-speed wireless data transfer. The term Wi-Max comes from Wireless (Wi) Microwave Access (MA). Like
802.11a, it uses the OFDM to achieve high data rate. It supports data rate up to
70 Mbps and distance up to 30 miles. 802.16 standard is for stationary user only, the mobility support will be added through 802.16e. These standards are still in development.
IEEE 802.20, also known as Mobile-Fi, is a standard designed from the ground up as a mobile system. It supports data rate up to 1.5 Mbps at high speed of 75 mph.
Sensor networks are special kinds of ad hoc networks. The sensor devices are low power, low cost. They are deployed in large quantities in areas that may not be easily accessible. Power saving, network scalability and self-healing of networks are some of the important topics in sensor networks.
Wireless LAN switches and routers are developed to connect the wireless net- works to the wireline backbone. Software-defined radio can be deployed to let the mobile station work with various types of wireless network. IP multimedia subsys- tem (IMS) allows the integration and roaming between wireline, 2G/3G/4G cellular,
VOIP, 802.11 and other network systems.
5 Combining all these technologies combined, the future generation of networks will be IP-centric converged networks. The services offered and applications provided on these networks could greatly change the way people communicate and collaborate with each other. The life style of each of us may be changed greatly by these networks.
Our research is part of effort to make that vision possible.
1.4 Motivation
802.11 ad hoc networks have many advantages over networks in infrastructure mode. Each station in an IBSS is a peer to each other. There is no central point of failure. The IBSS is set up automatically when the stations enter into their trans- mission ranges. There is no configuration and administration cost. In the battlefield scenario, ad hoc networks are the only feasible solutions. Infrastructure is hard to obtain and it takes time and resources to protect these network infrastructures. Ad hoc networks are more robust. An IBSS is still functioning even if a large number of stations are destroyed. For infrastructure BSS, the network is dead if the access point is damaged.
However, most of the deployments of 802.11 networks are still infrastructure BSSs.
The reason is that there are still a lot challenging issues in 802.11-based ad hoc networks. In this dissertation, we will study two of the fundamental issues. The first one is clock synchronization and the second is the construction of dominating set.
1.4.1 Clock Synchronization
Timing synchronization is always critical in network systems. Network synchro- nization has been gaining importance in the last 30 years when the transmission and switching of telecommunication systems turned digital. Network systems like
6 SDH (Synchronous Digital Hierachy)/SONET (Synchronous Optical Network) re- quire strict clock synchronization. Cellular mobile phone systems like GSM, North
American TDMA, CDMA and UMTS (Universal Mobile Telecommunications Sys- tem) all demand accurate time synchronization. Time synchronization has been proven to affect the quality of service for these systems[7].
In cellular phone networks, Global Positioning System (GPS) is utilized to syn- chronize base station’s clock. It is widely used in CDMA-based systems. Since major
3G standards like CDMA2000 and UMTS are all CDMA-based, GPS receivers are widely deployed in these systems. In addition to the GPS receiver, base station normally has one or two additional clock reference sources in case the GPS receiver fails to receive timing information from the satellite systems. It may have one clock source based on atomic frequency standards such as Rubidium standard, Caesium- beam standard and Hydrogen MASTER standard. It may also have an additional backup clock source based on quartz oscillator.
In 802.11 systems, we do not have a lot of options for clock synchronization. Due to the mass-market nature of the networks, only low cost hardware can be selected. In most of the 802.11 networks, crystal oscillators are deployed. They are much cheaper than the GPS systems and the atomic clocks, but the accuracy of the clock is much worse. This makes the clock synchronization problem more challenging.
1.4.2 Dominating Set Construction
A dominating set is a subset of vertices in which each vertex is either in the dominating set or adjacent to some vertex in the dominating set. If vertices in the dominating set form a connected graph, then the dominating set is called connected
7 dominating set. A dominating set with minimum size has many applications in wire- less ad hoc networks. A (connected) minimum dominating set is commonly used as the virtual backbone of the network [9]. It has been found extremely useful in routing
[13] [47] [48] [46], broadcast [11] [41] [10], and collision avoidance [45]. It is one of the fundamental issues in wireless ad hoc networks. However the problem is very hard to solve, minimum dominating set is shown to be NP-hard. In this dissertation, we will work on some efficient algorithms to find near optimal dominating set in wireless ad hoc networks.
1.5 Contributions
The scalability problem of the 802.11 timing synchronization function was first discovered in[19]. Unless the number of stations in an IBSS is very small, there is a non-negligible probability that stations may get out of synchronization. The more stations, the higher probability of asynchronism. In this sense, the current IEEE
802.11’s synchronization mechanism does not scale well; it cannot support a large- scale ad hoc network, say, from 150 to 200 stations.
We first study the ATSP protocol proposed in [19] to address the scalability prob- lem. ATSP works much better for large ad hoc networks than the IEEE 802.11 TSF standard. But it suffers two problems: the maximum clock difference between the fastest station and the slowest station may shoot up to over 500 µs for some occasions even when the hosts are stationary, the clock may be out of synchronization when the fastest station leaves the IBSS[23]. We analyze why ATSP suffer from these problems and then propose new remedies to improve the performance. We then proposed two
8 protocols: TATSF and ABTSF. TATSF stands for tiered adaptive timing synchro- nization function. TATSF is compatible with the 802.11 TSF and can control the maximum clock difference under 125µs. ABSTF stands for adaptive bidirectional timing synchronization function. ABTSF is deigned from scratch and it can achieve even better synchronization. The maximum clock difference is under 50µs[51].
By careful examination of the clock models and reasons for clock adjustment errors, we are able to design a compatible protocol called SATSF[52]. SATSF stands for self-adjusting timing synchronization function. SATSF meets all the desirable requirements: simple, scalable, accurate, compatible, bounded and adaptive to station mobility. For SATSF, the maximum clock offset can be controlled within 20 µs with the worst clocks allowed by the standard. In particular, we want to point out that
SATSF is completely compatible with the current 802.11 TSF protocol by keeping the beacon frame format intact. Since beacon frame is one of the most important control frames in 802.11 protocols, a received beacon is validated by the current protocols.
A format change may fail these validations and make the TSF protocol fail in a mixed-mode IBSS.
The challenge of our work is to meet the industry expectation of time accuracy
(maximum clock offset under 25µs) with a compatible, scalable, mobility-friendly and cheap solution. It is not difficult to satisfy some of the requirements. But it is quite an effort to hit all the targets. SATSF can meet all these challenges.
After getting satisfying synchronization accuracy in a single-hop IBSS, we expand our work to address the scalability problem in an 802.11-based multihop mobile ad hoc network (MANET). The single-hop solutions cannot be adopted directly into the MANET. Simply giving priority to the fastest station(s) will not be enough. It
9 may take very long time for the timing information of the fastest station to reach the stations many hops away. The accuracy of these types of TSFs will suffer in multihop networks. We propose a new protocol called multihop adaptive timing synchronization function (MATSF). For MATSF, the maximum clock difference can be controlled within 50 µs with the worst clocks allowed by the standard[53]. The improvement is more than 400% over the current solutions of similar complexity.
The protocol converges in less than 10 seconds. It converges several times faster than current solutions.
We study the problem of connected dominating set construction. We present a
MAC-Layer Timer-based Connected Dominating Set Construction Protocol (MTCDS).
There are many advantages by constructing dominating set in the MAC layer. It can detect topology changes quickly; the resulting dominating set can be utilized by up- per layer protocols such as routing protocols. Most of the information needed by the protocol is already available in the underline time synchronization protocol. There is little or no overhead by piggybacking on the existing TSF.
MTCDS is simple, distributed, inexpensive, and adaptive to station mobility.
Only one hop neighbor information is needed. The simulation results show that our protocol can construct connected dominating set using 35% to 60% less nodes than other distributed connected dominating set protocols.
MTCDS can be easily extended to execute greedy algorithms in ad hoc networks.
We are able to achieve very good results by extending the framework to the construc- tion of power aware CDS and efficient broadcasting schemes.
10 1.6 Organization of this Dissertation
The rest of this Dissertation is organized as follows.
Chapter 2 gives a brief overview of clock synchronization history. In particular, we introduce the clock synchronization protocol and the beacon frame format defined by the IEEE 802.11 TSF. A mathematical model for clocks used by the common network systems is presented. We also examine various clock technologies and their performance characteristics.
Chapter 3 introduces the issues and the challenges for the two fundamental prob- lems we will study in this dissertation. We first show the importance of clock syn- chronization in wireless ad hoc networks and then list the problems of the current
802.11 TSF. The most important issue is the scalability issue. We follow up with the requirements for ideal clock synchronization protocols in 802.11 networks. We give the formal definition of the minimum dominating set (MDS) problem including the connected dominating set problem. The significance of the minimum dominating set and the applications of the MDS protocols are outlined. We conclude the chapter with the investigation of the challenges of minimum dominating set construction in wireless ad hoc networks.
Chapter 4 reviews the analytical results of the scalability issue of 802.11 Timing
Synchronization Function. The probability of at least one successful beacon trans- mission during the beacon window and the probability of one particular station’s successful beacon transmission are presented. We then define the metrics for the asynchronisim. We further analyze the root causes of the asynchronism to set up the foundations of our proposed new solutions in later chapters.
11 In chapter 5, we first overview the related clock synchronization protocols includ- ing logical clock synchronization, network time synchronization used for the Internet and various clock synchronization protocols applied in sensor networks. These proto- cols cannot be adapted easily into the 802.11 TSF frameworks. We then investigate the TSF protocols designed for single-hop 802.11 IBSSs. We can see ATSP effectively solves the scalability problem, but cannot handle mobility very well. ABTSF and
TATSF can handle both scalability and mobility very well. However the maximum clock offset is still above the industry expectation of 25µs. Finally we introduce
SATSF, which is a self adjustment protocol with bounded frequency adjustment.
The performance of SATSF is very well and exceeds the industry requirements. More important, the protocol is completely compatible with the current 802.11 TSF. It is ready to be deployed in a stand alone or mixed environment. Extensive simula- tion works are done to present the maximum and average clock offsets under various conditions.
In chapter 6, we study the clock synchronization problem in 802.11-based multihop ad hoc networks. First we overview the current solutions in the field, then we propose a new protocol called MATSF. MATSF can achieve very good clock synchronization accuracy through beacon transmission prioritization, bounded frequency adjustment and construction of dominating set. Simulation shows that MATSF can improve the maximum clock offset by at least 400%.
Chapter 7 studies the construction of dominating sets in 802.11-based multihop mobile ad hoc networks. We first examine the importance and the applications of the dominating set problem in wireless ad hoc networks. We then proceed to review the current solutions for connected dominating set construction in wireless ad hoc
12 networks. The current solutions suffer from one or more problems: the resulting
CDS (Connected Dominating Set) is not very small; the protocols may not adapt to topology changes; the protocol introduces extensive overheads. All these issues are undesirable for the construction of dominating set in mobile ad hoc networks; there is a need to find a better solution. We are able to find such a solution to resolve all these problems. The protocol is called MTCDS. We prove that structure we build is a dominating set and it is connected. We analyze the handling of station mobility by MTCDS. The design and implementation issues are studied. We compare the performance of MTCDS with the current solutions. MTCDS constantly out-performs the current solutions in term of the size of dominating set. MTCDS can be extended to be a framework for the implementation of greedy algorithm in mobile ad hoc networks. We are able to construct an energy-aware connected dominating set with small size and the resulting network has a longer life span than the current solutions.
The framework can be applied to optimize other facts of mobile ad hoc networks in a simple and efficient way.
Chapter 8 summarizes the contributions of the dissertation, and gives pointers to future researchs that can be built up on top of the current research achievements.
Appendix A compiles a list of abbreviations used in the dissertation.
13 CHAPTER 2
BACKGROUND
2.1 History of clock synchronization
Clock synchronization is a classic issue in telecommunication networks for the past
30 years. Digital switching equipments (e.g. 5ESS or DMS) require synchronization to avoid slips at the input elastic stores. Voice traffics are not affected much by the synchronization slips. However, the slips have significant impact on circuit switched data services. Synchronization became to gain importance with the introduction of ISDN (Integrated Service Digital Network). The needs to get accurate reference clock have grown with the wide deployment of SDH and SONET. Currently most of the communication systems like ATM, GSM, CDMA and UMTS require access to common timing references.
Clock synchronization is not only needed in telecommunication networks; synchro- nization of large number of oscillators can be found in nature. Some types of fireflies
flash their light organs at regular but independent intervals if they are far away.
When they gather in a close proximity, they start to synchronize their light organs until they flash in unison[27]. Another example is the synchronization of individual
fibers in heart muscles to produce a regular heartbeat.
14