March 2001 doc IEEE 802 11 01154 OFDM

March 2001 doc. : IEEE 802. 11 -01/154 OFDM as a High Rate Extension to the CCK-based 802. 11 b Standard Steve Halford, Ph. D. Mark Webster Jim Zyren Intersil Corporation Submission 1 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 Why OFDM for High Rate? • OFDM recognized as best solution for W-LAN üSelected by 802. 11 a & ETSI for W-LAN at 5 GHz • Intersil’s proposed OFDM waveform offers: üFully backwards compatible with 802. 11 b üProvides forward compatibility with 802. 11 a üMeets data rate needs & expectations set by 802. 11 a üBest complexity versus performance trade üGood performance in real world W-LAN üWell-known & proven technology Submission 2 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 Overview of Intersil’s Proposal for 802. 11 g Submission 3 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 OFDM for 2. 4 Ghz band • Use long & short preamble for backward compatibility – Ultra-short preamble possible for certain CCA modes • Replace current CCK data with OFDM – Data modulation identical to 802. 11 a • Maintain the same 2. 4 GHz channels – 25 MHz center frequency spacing (wider than 802. 11 a) • Use 802. 11 a clock rates (20 MHz) for OFDM mode – Data rates identical to 802. 11 a (6, 9, 12, 18, 24, 36, 48, 54 mbps) – Originally proposed using 802. 11 b clock of 22 MHz – Now feel acceptance would be faster with 802. 11 a rates • Technical differences are very small • Rate change circuitry is common in low power IC Submission S. Halford, M. Webster, & J. Zyren, Intersil Corporation – Open to changes from the 4 group

March 2001 doc. : IEEE 802. 11 -01/154 Packet Structure: Backwards Compatible 802. 11 g LONG (Short) Preamble Packets PREAMBLE/HEADER (Barker Words -- 802. 11 b) 192 usecs (Long) 96 usecs (Short) Signal Extension PSDU SELECTABLE OFDM Symbols OFDM SYNC @ 6, 9, 12, 18, 24, 36, 48 or 54 Mbps 12 usecs 6 usecs Uses OFDM Modulation n Existing. 11 b radios will recognize preamble and header § § n n Length field will be correctly decoded Use reserve bits in header to denote switch Add OFDM Sync after. 11 b header to simplify design Add signal extension for SIFS compatibility OFDM Proposal is compatible with 802. 11 b Submission 5 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 OFDM Symbol Structure • OFDM uses industry standard R=1/2, K=7 code – Known performance, complexity, and IP issues • OFDM symbols are formed by IFFT of symbol block – Maps the coded data onto narrow carriers – IFFT block includes 4 pilot/training signals – Carriers retain orthogonality in multipath • OFDM symbols include guard intervals for multipath – Provides “buffer” to absorb ISI Preceding Symbol Multipath will cause preceding symbol to “bleed” into current symbol. Guard interval absorbs this interference Submission Guard Interval 16 Samples 6 4 usecs OFDM Symbol 64 pt. IFFT of coded data 64 Samples time S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 Radio Design Issues • Requires a change in baseband processor only – Current RF gives adequate performance up to 36 Mbps • OFDM preserves current channelization – 3 channels spaced by 25 MHz (U. S. deployments) • Power requirements are same as present products • For 48 Mbps & 54 Mbps, new RF design is required – Standard would be in place & spur development – Design issues are well understood Submission 7 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 Performance of OFDM with Prism II Radio Implementation loss due to radio (Loss relative to ideal OFDM performance) 36 Mbps Mode** • 10% PER -- 2. 7 d. B • 1% PER -- 4. 1 d. B 24 Mbps Mode** • 10% PER -- 1. 5 d. B • 1% PER -- 2. 1 d. B Current radio provides sufficient quality to operate at rates up to 36 Mbps ** see notes pages for more details Submission 8 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 Preambles and Throughput Submission 9 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 Throughput Impact of OFDM Sync • Proposed OFDM Sync in addition to 802. 11 b & SIFS pad – Up to 18 usecs of additional overhead – Simplifies the receiver design – Allows future flexibility • What is the impact on throughput? Decrease in throughput 100 byte: 310 kbits/sec 1000 byte: 500 kbits/sec 2346 byte: 704 kbits/sec Throughput impact is negligible Submission 10 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 Complexity and Performance for 802. 11 g Submission 11 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 OFDM Transmitter doc. : IEEE 802. 11 -01/154 • OFDM distributes “equalization” between the transmitter & receiver – Single carrier proposals relies on receiver for multipath protection – W-LAN systems are in receive mode 90% of time so reducing receive complexity is critical for power savings • OFDM adds IFFT and cyclic extension operations to Only added items compared to single transmitter carrier system like PBCC – Simplifies the equalizer in the receiver Uncoded Information Bits Data Scrambler Convolutional Encoder Puncture To RF Transmitter Submission Interleave 12 Constellation Mapping (bits to symbols) 64 -pt Inverse FFT Cyclic Extension S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 OFDM Receiver Structure • Major difference is use of FFT to simplify equalizer • Reduce tracking complexity with pilot Frequency Domain Equalizer: tones Multiply each tone by inverse gain & Receiver From A-to-D CNCO phase of the channel Timing Adjust Trim Guard Interval FFT &FEQ 52 tones Soft-Decisions on Bits (symbol to bits) Extract 4 Pilot Tones Carrier/Timing Correction Frequency Correction To MAC Submission De-interleave & De-puncture Viterbi Decoder De-Scrambler 13 Compute Branch Matrix S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 Error Correction Coding for High Rate • Encoding process is relatively low complexity • Decoding complexity depends on code properties • Decoders are based on Viterbi algorithm • VA searches trellis at each step for most likely state sequence • Complexity depends on the number of states in decoder • Number of states determines size of the trellis searched by VA • PBCC-11 & OFDM use a 64 -state decoder • PBCC-22 uses a 256 -state decoder • Trellis size is 4 x the equivalent all OFDM decoders • Trace-back depth is larger than OFDM-24 OFDM has a less complex error correction code Submission 14 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 PBCC 256 state code vs. Industry Standard 0. 8 d. B Advantage for PBCC-22 at 1% PER 1. 0 d. B Advantage for PBCC-22 at 10% PER 0. 8 -1. 0 d. B Coding Gain relative to punctured industry standard code. Requires Trellix 4 x larger. . . If AWGN performance is needed, better codes could be developed for OFDM Is 1 d. B worth greatly increased complexity? Submission 15 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 Multipath & Equalization for 802. 11 g Submission 16 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 Performance & Complexity Trades • W-LAN performance is dominated by multipath • OFDM is designed for both AWGN and multipath – Error correcting code to provide AWGN – Use guard interval to absorb ISI (0. 96 d. B AWGN loss) – Use pilot tones for improved tracking (0. 34 d. B AWGN loss) • PBCC is optimized for AWGN only – Error correcting for AWGN OFDM is lesscode complex than PBCC – Multipath depends entirely on forperformance W-LAN environment receiver Submission 17 S. Halford, M. Webster, & J. Zyren, Intersil Corporation – Tracking depends entirely on receiver

March 2001 doc. : IEEE 802. 11 -01/154 Equalizers for PBCC • Linear Equalizer -- Invert the channel with linear filter – Length of filter depends on number of multipath rays (15 - 20 taps) – Matrix Inverse required for each packet – More complex than FFT based equalizer for OFDM • Decision Feedback Equalizer (DFE) -- Subtracts interference – Uses hard decisions on received symbols prior to error correction – May need a whitened matched filter (matrix inverse to compute) • Viterbi Equalizer – – Maximum likelihood sequence estimate or MLSE Performance depends on number of paths “tracked” May require whitened matched filter (# of taps ? ) Finds the most likely sequence of transmitted symbol based on channel Neither Linear nor DFE& equalizers make sense for • Similar complexity implementation to decoding a PBCC convolutional code Submission 18 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 MLSE/Viterbi Equalizer Received Data Whitened Matched Filter Viterbi Equalizer Equalized Symbols • Equalizer estimates the most likely sequence based on knowledge of the channel and the received data – Viterbi itself requires only a channel estimate – Matrix inverse may be required for WMF • Can include in Viterbi -- affects the observed channel • Similar to decoding a convolutional code – Searches a trellis for best path between states • MLSE is the likely equalizer for PBCC-11 & 22 – Need to track 4 or more paths for adequate 19 S. Halford, M. Webster, performance & J. Zyren, Intersil Corporation Submission

March 2001 doc. : IEEE 802. 11 -01/154 MLSE: Complexity Considerations • Complexity is similar to convolutional decoder • Number of states depends on constellation and number Numbersize of States in Joint Decoder of multipath rays being tracked Example Track 4 rays for 8 -level PSK (PBCC-22) Number of states = 83 = 512 states Eight times as complex as the 64 state PBCC 11/OFDM decoder& only 4 rays are being tracked! ** See pg. 590, J. G. Proakis, Digital Communication, 3 rd Ed. , Mc. Graw-Hill, 1995. Submission 20 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 Joint Decoder MLSE: Complexity Considerations • Possible to use single “super-trellis” for decoder • Includes both multipath and FEC memory Number of States in Super. Trellis = S ML-1 M: Constellation Size, L: Number of paths tracked, S: number states in FEC Example Track 4 rays for 8 -level PSK with 256 state Conv. Code (PBCC-22 or 33) ü Number of states = 256 x 83 = 217 = 131072 Submission 21 Over 2000 times as complex as the 64 state PBCC-11/OFDM decoder S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 OFDM gives MLSE type performance • OFDM uses a guard interval to absorb multipath interference • Outside the guard interval, signal is multipath free – Multipath causes individual tones to fade • After FFT, each tone is multipath free – Relative fade is known from channel estimation • Viterbi Decoder of error correction code gives MLSE in multipath – Reliability of each soft-decision is weighted by known fade – Optimum receiver is realized with only a FFT Submission 22 Halford, M. Webster, & J. Zyren, Intersil Corporation – True provided multipath is S. entirely inside guard

March 2001 doc. : IEEE 802. 11 -01/154 OFDM: MLSE performance w/o complexity OFDM Multipath Tolerance • OFDM proposal includes 800 n. Secs Guard Interval • Equivalent to 800 e-9 x 11 e 6 = 8. 8 paths at PBCC symbol rate • Multipath tolerance equivalent to tracking 8 paths • FFT complexity is approximately twice the complexity of a 64 state decoder Equivalent SC MLSE Complexity This is 215 (over 32, 000) times the complexity of the 64 state decoder! Submission 23 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 Impact of interference on 802. 11 g Submission 24 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 Interference in 2. 4 GHz band • 2. 4 GHz spectrum is a shared resource – Blue. Tooth & other FH systems generate in-band interference on 802. 11 b & 802. 11 g radios – Other sources of interference include microwave ovens • Higher data rates specified by 802. 11 g will be more sensitive to interference – Errors generated by presence of interference source can greatly influence throughput • Blue. Tooth enabled devices will proliferate at same time as 802. 11 g PBCC is sensitive to real world interference sources Submission 25 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 PBCC performance is sensitive to Blue. Tooth E. Zehavi, et al (IEEE documents IEEE 802. 1101/061 r 0 & IEEE 802. 15 -01/066 r 0) showed that the throughput of coded 8 -PSK w/o an interleaver was very sensitive to the presence of a Blue. Tooth- like interferer. Submission 26 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 Extending PBCC to higher rates (>22 Mbps) Submission 27 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 Approaches to Higher Data Rates • OFDM provides a known path to higher rates • Higher data rates can be achieved by: – Increasing the constellation size and/or decrease code rate • Used by OFDM to give rates of 6 Mbps to 54 Mbps • PBCC-22 uses 8 -psk with rate 2/3 code to go from 11 Mbps (QPSK with rate 1/2) to 22 Mbps – Increasing symbol rate • PBCC-33 uses 1. 5 x clock speed to go from 22 Mbps to 33 Mbps • Increasing the data rate increases the required SNR OFDM equalizer complexity is same for all rates -What is the impact on the PBCC receiver? Submission 28 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 Are Higher Data Rates Possible? • OFDM Equalizer has fixed complexity for all proposed rates – Higher rates does impact performance due to fading of tones • Guard interval however reduces the impact independent of rate • MLSE complexity will grow exponentially when constellation size increases – Higher rates will impact performance • No guard interval to protect from increased ISI sensitivity – Example: Track 4 paths -- Number of states = (constellation size)4 -1 • 22 Mbps (8 -PSK) requires 83 = 512 states (8 x the PBCC-11 decoder) 3 = 4096 states Extending to higher constellation • 33 PBCC Mbps (16 -QAM) will rates requireby 16 increasing (64 x the PBCC-11 decoder) is not practical • 44 Mbps (64 -QAM) will require 643 = 262144 states (4096 x the Submission 29 S. Halford, M. Webster, & J. Zyren, Intersil Corporation PBCC-11 decoder)

March 2001 doc. : IEEE 802. 11 -01/154 Are Higher Data Rates Possible? • OFDM uses a fixed symbol rate for all data rates – Guard interval protection is same for all rates • PBCC-33 is PBCC-22 at a higher symbol rate – Pulse shaping used to keep same spectral width • Increasing symbol rate impacts performance – Increasing timing accuracy requirements • Increasing rate increase number of equalizer paths – Example: 8 -PSK -- Number of states = 8(number of paths -1) • 22 Mbps (11 Mhz, 4 paths) -- 84 -1 = 512 states (8 x the PBCC-11 decoder) Extending PBCC to higher increasing • 33 Mbps (16. 5 Mhz, 6 paths) rates -- 86 -1 = by 32, 768 states (512 xsymbol PBCC-11 rate decoder) is not practical states PBCC-11 Submission • 44 Mbps (22 Mhz, 8 paths) -- 88 -1 30 = 2, 097, 152 S. Halford, M. Webster, (32, 768 x & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 Conclusions on OFDM for 802. 11 g Submission 31 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 doc. : IEEE 802. 11 -01/154 Summary of Data Rates & Parameters Submission 32 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 Conclusions doc. : IEEE 802. 11 -01/154 • OFDM is forward & backwards compatible – Uses existing long & short preamble for compatibility’ – 802. 11 a signaling used in place of CCK – Minor impact on throughput of added headers • OFDM offers the highest rates of all proposals – 36 Mpbs with current radio (baseband only change) – 48 & 54 Mbps possible with new radio design – PBCC complexity grows exponentially Submission 33 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 Conclusions doc. : IEEE 802. 11 -01/154 OFDM is ideal for W-LAN environment – Equalization split between transmitter & receiver for lower overall complexity – Lower complexity error correction code – Nearly MLSE without complexity – PBCC Joint Decoder approach requires RSSE • Complexity vs. Performance ? OFDM is robust to narrowband interference – PBCC seems to have an inherent problem with BT Submission 34 S. Halford, M. Webster, & J. Zyren, Intersil Corporation

March 2001 Conclusions doc. : IEEE 802. 11 -01/154 OFDM will meet regulatory approval – All high rate waveforms possible under new rules (? ) – OFDM will be in this band -- IEEE should ensure network compatibility • OFDM has been developed in an open process – No hidden details – Complexity of PBCC never adequately described – Complexity and design is well known & proven Submission 35 S. Halford, M. Webster, & J. Zyren, Intersil Corporation