ENGG*6090 Reconfigurable Computing System

Topic Review Presentation April 3rd, 2006

Accelerating WiMAX System Designs with FPGAs

[Xiaoguang (Shaw) LI][0190094]

Instructor: Professor Shawki Areibi Reviewed Papers:

1. WiMAX Opportunities and Challenges in a World, White Paper developed for the CDMA Development Group, July 2005. 2. IEEE 802.16a Standard and WiMAX Igniting Wireless Access, Worldwide Interoperability for Access Fourm. 3. Telephony’s Complete Guide to WiMAX, WiMAX © Telephony 2004 4. OFDM for wireless communications systems/Ramjee Prasad., p.cm – {Artech House universal personal communications series}, ISBN 1-58053-796-0 5. The principles of OFDM, RF signal processing, January 2001 6. Joaquin Garcia and Rene Cumplido, On the design of an FPGA-Based OFDM modulator for IEEE 802.16-2004, Proceedings of the 2005 International Conference on Reconfigurable Computing and FPGAs, Puebla City, Mexico 7. Maryse Wouters and et al., Real Time Implementation on FPGA of an OFDM based Wireless LAN extended with Adaptive Loading, Imec vzw, DESICS, Kapeldreef 75, B-3001 Heverlee, Belgium. 8. Accelerating WiMAX System Design with FPGAs, White Paper, Altera, Oct. 2004 9. http://www.xilinx.com/esp/wireless/bfwa/ieee_802_16.htm 10. Kwang-Cheng Chen et al., A Programmable Architecture for OFDM-CDMA, IEEE Communications Maganine, Nov. 1999 2 Outline •Introduction •WiMAX and OFDM •FPGA-based WiMAX system designs •Design Challenges/Objectives •WiMAX solutions with Xilinx FPGAs •WiMAX solutions with Altera FPGAs •A real design of an FPGA-Based OFDM modulator •Results •Conclusion

3 Convergence in Wireless Communication

CDMA2000 CDMA W-CDMA GSM CDMA HSDPA TDMA GPRS EDGE Cellular 2. 3.5G 3.75G Evolution Convergence 802.16e

802.16d 802.16a 802.16g 802.11b Wireless MAN 802.11g Wireless Proprietary Network Wireless LAN Evolution

1999 2003 2005 2006 Time Introduction 4 Wi-Fi (Wireless Fidelity)

•Short for Wireless Fidelity and is meant to be used generically when referring to any type of 802.11 network, whether 802.11b, 802.11a, 802.11g, dual-band, etc. Disadvantages: •Security- greater exposure to risks •Speed – Slower than cable •802.11b: 1 to 11 Mbps •802.11g: run at 54 Mbps, but realistically about 20-25 Mbps and about 14 Mbps when b associated. •Range – Affected by various medium •802.11a: Indoor 40 to 300 feet; outdoor 100 to 1000 feet •802.11b: Indoor 100 to 300feet; outdoor 400 to 1500 feet

Introduction 5 IEEEIEEE 802.16802.16 StandardsStandards (WiMAX (Worldwide Interoperability for Microwave Access) is a trade name of a group of IEEE 802.16 Standards) 802.16 802.16a/REVd 802.16e Completed Dec. 2001 802.16a: Jan. 2003 Dec. 2005 802.16 REVd: Q3 2004 Spectrum 10 to 66 GHz <11 GHz <6GHz Channel Line-of-Sight only Non line-of-sight Non line-of-sight Conditions 32 to 134 Mb/s at Up to 75 Mb/s at 20 MHz Up to 15 Mb/s at 28 MHz channelization 5 MHz channelization channelization Modulation QPSK, 16QAM OFDM 256, OFDMA 64 QAM, Same as ERVd and 64 QAM 16QAM, QPSK,BPSK Mobility Fixed Fixed and Portable Mobility, Regional Roaming Channel 20, 25 and 28 MHz Selectable channel bandwidths Same as REVd Bandwidths between 1.25 and 20 MHz, with up to 16 logical sub-channels Typical Cell 1 to 3 miles 3 to 5 miles; Maximum range 30 1 to 3 miles Radius miles based on tower height, antenna gain and transmit power 6 WiMAX Milestones & Roadmap

Introduction 7 Single Carrier Modulation – Time Domain

Signal

time

Carrier

time

Introduction 8 Single Carrier Modulation – Frequency Domain

Unmodulated

Carrier Signal

frequency Baseband

Modulated Carrier Signal

frequency Baseband Introduction Baseband 9 Problem: Data transmission over multiple channels

•In classical terrestrial broadcasting scenario, we have to deal with a multiple-channel: the transmitted signal arrives at the receiver in various paths of different length. •Since multiple versions of the signal interfere with each (inter symbol interference (ISI)), it becomes very hard to extract the original information.

•Channel impulse response: the signal at the receiver f a single pulse is transmitted. •A received symbol can theoretically be

influenced by τ max / T previous symbols. •This influence has to be estimated and compensated for in the receiver. Introduction 10 Single Carrier Approach

A scenario that is characterized by the following conditions:

1 •Transmission Rate: R = = 7.4Msym / s T •Maximum channel delay: τ max = 224µS

τ For the single carrier system, this results in an ISI of: max ≈1600 T The complexity involved in removing this interference in the receiver is tremendous. In the scenario under consideration here, using an approach will only lead to sub-optimal results. This is the main reason why the multi carrier approach becomes so popular.

Introduction 11 Multi Carrier Approach

•The original data stream of rate R is multiplexed into N parallel data streams of rate R mc = 1 / T mc = R / N each of the data streams is modulated with a different frequency and the results signal s are transmitted together in the same band. τ τ The ISI for each sub system reduces to max = max τ T N •T •If N=8192, ISI is max = 0.2 mc Tmc •Such little ISI can often be tolerated and no extra counter measure such as an equalizer is needed. However, the complexity of a receiver with 8192 parallel paths still isn’t feasible. This is why OFDM comes to the picture.

Introduction 12 of OFDM Signal

Introduction 13 Design Challenges

•Performance (Processing Speed)

•WiMAX requires higher throughput and data rate than those in cellular systems. •WiMAX utilizes advanced signal processing techniques, such as Turbo coding/decoding, FFT/IFFT, beam forming, MIMO, CFR, and PDP which are very computationally intensive and require billion MAC per second. •WiMAX requires hardware platform that must have significant processing abilities. •Flexibility (Reprogrammability) •WiMAX standards are still evolving. •Interoperability and compliance testing are in the early stages. •Requirement to adapt systems: Throughout the development cycle including after deployment in the field.

FPGA-based WiMAX System Designs 14 Design Challenges cont. •Time-to-Market •Many new players have emerged. •WiMAX is an emerging technology, time-to-market is a key differentiator for OEMs looking for early success gaining market share. •This has a direct effect on the development cycle and choice of hardware platform, with designers requiring easy-to-use development tools, software, boards, and off-the-shelf IP and reference designs in order to accelerate the system design. •Low cost •Need low cost points to help drive rapid adoption of technology. •Need a way to reduce costs once standards mature and when loss of small degree of flexibility is acceptable (Structured ASIC). •Power Dissipation •Despite the demanding performance requirements power dissipation must be as low as possible. FPGA-based WiMAX System Designs 15 Technology Comparison

Technology Performance Cost Power Flexibility Memory BW I/O BW

GPP LOW LOW HIGH HIGH LOW LOW

PDSP Medium Medium Medium Medium Medium LOW

ASIC HIGH HIGH LOW LOW HIGH HIGH

FPGA Med-High LOW Low-Medium HIGH HIGH HIGH

FPGA-based WiMAX System Designs 16 •Virtex-4 FPGAs (LX, FX, and SX)

LX: High Performance Logic •Highest Logic-to-Feature ratio •Highest I/O-to-Feature ratio •SX: Ultra-high-performance signal processing •Highest DSP-to-Feature ratio •Highest Memory-to-Feature ratio •FX: Embedded processing and high-speed serial connectivity •Embedded IBM PowerPC processor and MAC •Rocket IO multi-gigabit serial transceivers

FPGA-based WiMAX System Designs 17 Embedded DSP Blocks

¾Most powerful embedded DSP capability in FPGA industry ¾Pipeline registers enable 500 MHz performance ¾Cascade logic enables sustained 500 MHz performance throughout DSP column ¾Build high-speed multi-level filters using DSP slices

FPGA-based WiMAX System Designs 18 Virtex-4 FX: Immersed PowerPC with APU

•New Auxiliary Processing Unit (APU) •Direct interface from CPU pipeline to FPGA logic •Simplifies integration of Coprocessor an hardware accelerator •APU Acceleration •Reduce number of bus cycles by factor of 10X •Increase performance by over 20X

FPGA-based WiMAX System Designs 19 Off-the-Shelf IPS for WiMAX

FPGA-based WiMAX System Designs 20 Xilinx DSP Design Tools and Flow

DSP System Simulation Real-time Debug (using ChipScope)

MATLAB/Simulink HDL Co-Simulation ModelSim ISE PCI, JTAG

Implementation Simulate HDL Hardware Modules in-the-loop

FPGA-based WiMAX System Designs 21 Flexibility Feature from Altera

•Additional protocol support is required to ensure compatibility with future products. •Enhancements or bug fixes are necessary.

•Send an update from the development location through a network to the Stratix II device. •Store the update in the memory. •Update the Stratix II device with the new data.

FPGA-based WiMAX System Designs 22 Altera’s Low Cost Solution: Structured ASIC

Structured ASICs leverage standard- cell technology and the most advanced semiconductor processes to embed logic and hard functions— such as memory, phase-locked loops (PLLs), clock networks, and power bussing—into pre-engineered, pre- verified base layers. The structured ASIC is customized using just the top metal layers. The result is a device that meets today’s performance requirements for advanced systems in much less time and at a much lower NRE costs.

FPGA-based WiMAX System Designs 23 Altera’s Structured ASIC Cont.

The 90-nm HardCopy ® II family is built on an array of fine-grained on average 40 percent structured cells, called HCells, that less power and 40% delivers the density, low cost, and faster than their high performance required for high- equivalent FPGAs volume advanced systems.

HCells are grouped into HCell macros to implement a portion of a Stratix® II adaptive logic module (ALM) or a section of a digital signal processing (DSP) block. Altera’s Quartus® II design software maintains a library containing a pre-verified, pre- characterized HCell macro for every ALM configuration. Quartus II software uses the library to map each Stratix II ALM into the HardCopy II design.

FPGA-based WiMAX System Designs 24 Altera’s Structured ASIC Cont.

25 Transmitter Block Diagram for the OFDM PHY.

Joaquin Garcia and Rene Cumplido, On the design of an FPGA-Based OFDM modulator for IEEE 802.16-2004, Proceedings of the 2005 International Conference on Reconfigurable Computing and FPGAs, Puebla City, Mexico

FEC (Forward Interleaving/ Guard Error Correction) Mapping IFFT Interval (Cyclic In Phase and Prefix) RF Quadrature Modulation Modulation FPGA-based WiMAX System Designs 26 System Modeling in Simulink

IFFT and Prefix adding Model

Interleaver Model

PSK or QAM Mapping

FPGA-based WiMAX System Designs 27 Results

FPGA-based WiMAX System Designs 28 Conclusion

•WiMAX is an emerging technology with significant potential and is poised to revolutionize the “broadband wireless ” market. •The diverse hardware requirements including processing speed, flexibility, integration and time-to market necessitate an FPGA based implementation platform. •Xilinx and Altera all provide WiMAX solutions including Libraries, Reference designs, and comprehensive solutions from AllianceCore partners.

Conclusion 29 30 Back up pages

31 WiMAX Solutions with Altera FPGAs

•Startix II FPGAs

I/O Channels with External Memory Adaptive Logic Interface Circuitry Modules

Digital Signal M512 Block Processing (DSP) Block M4K Block M-RAM Block High-Speed I/O Channels With Dynamic Phase Phase- Alignment (DPA) Locked Loop Startix II EP2S60 (PLL) 32 WiMAX Solutions with Altera FPGAs

Orders of magnitude higher than single-chip DSP •Support a variety of chip available in the marketplace today! multiplier bit sizes (9*9, 18*18, 36*36) •Support different operation modes (multiplication, complex multiplication, multiply-accumulation, and multiply-addition) •Offer a DSP throughput of 2.8 giga multiply- accumulates per second (GMACS) per DSP block. •The largest Startix II device, the EP2S180 device, has 96 DSP blocks that offer a throughput of 284 GMACS and support

Stratix II DSP Block Architecture up to 384 18*18 multipliers.33 WiMAX Solutions with Altera FPGAs

What can help OEMs (Original Equipment Manufacturer ) decrease Time-to-Market when they developing WiMAX compliant products?

•Available off-the-shelf IPs and reference designs •Altera provides the RS (Reed- Solomon) Compiler and Viterbi •Easy-to-use development tools/software Compiler IPs that are fully •Availability of development boards parameterizable. The source code Design for FPGA is automatically created at the push of a button. •Altera provides FFT IP function that Matlab/Simulink VHDL is a high performance, highly- parameterizable FFT processor. Design Matlab/Simulink DSP Builder FPGA The FFT function implements a Verification FPGA Radix-2/4 decimation-in-frequency FFT algorithm for transformm lengths of 2 where 6

34 WiMAX Solutions with Altera FPGAs

System on a programmable chip (SOPC)

•Some functions in WiMAX system are better suited for implementation in software. •Altera provides the Nios II family of embedded processors consisting of fast, economy, and standard cores optimized for a specific price and performance range.

35 WiMAX Solutions with Altera FPGAs

Altera’s HardCopy® structured ASICs are unique because they embed hard functions from the Stratix® FPGA series (and equivalent I/O) into the base layers, delivering unprecedented design flexibility. HardCopy devices allow you to: •Use Altera's Quartus® II software along with the EDA tools of your choice to generate your design •Test your design in-system and at-speed with a Stratix or Stratix II FPGA •Migrate seamlessly from your FPGA design to a low cost, pin-compatible HardCopy device with no risk and very little effort •Switch back to FPGAs if you need to change the design to accommodate a new standard, customize for a specific market or application, or if production volumes decrease

36 Orthogonal Frequency Division Multiplexing (OFDM)

•Modulation: a mapping of the information on changes in the carrier phase, frequency or amplitude or combination. •Multiplexing: method of sharing a with other independent data channels. •OFDM is a combination of modulation and multiplexing. •In OFDM the signal itself is first split into independent channels, modulated by data and then re-multiplexed to create the OFDM carrier.

37 Orthogonal Frequency Division Multiplexing (OFDM)

Example: given a certain bit sequence, what is the development of the OFDM signal using 4 sub-carriers.

The first few bits are 1, 1, -1, -1, 1, 1, 1, -1, 1, -1, -1, -1, -1, 1, -1, -1, -1, 1, …

Serial to parallel conversion of data bit

38 Orthogonal Frequency Division Multiplexing (OFDM)

39 Orthogonal Frequency Division Multiplexing (OFDM)

40 Orthogonal Frequency Division Multiplexing (OFDM)

A simple OFDM system model:

Advantages: •Efficient use of available bandwidth since subchannels are overlapping. •Solve the multipath propagation problem.

41