Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs)

1. Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs) Submission Title: [Questions and Answers on the properties of the MB-OFDM PHY] Date Submitted: [ 16 September, 2003] Source: [Charles Razzell] Company [Philips Semiconductors] Address [1109 McKay Drive, San Jose, CA95131, California, USA] Voice:[+1 408-474-7243], FAX: [+1 408-474-5343], E-Mail:[charles.razzell@philips.com] and [Anand Dabak] Company [Texas Instruments, Inc.] Address [12500 TI Blvd, Dallas, TX75243, Texas, USA] Voice [214-480-4220 ], FAX:[972-761-6966 ], E-Mail:[dabak@ti.com ] Re: [IEEE P802.15-02/372r8 “IEEE P802.15 Alternate PHY Call For Proposals” dated 17 January, 2003] Abstract: [The following contribution is offered in response to assertions that have been made both formally and informally by 15.3a members. These assertions address the fundamental properties of the MB-OFDM signal which has been down-selected for confirmation as the alternate PHY for 802.15.3 wireless PANs. ] Purpose: [This document reflects the contents of 802.15-372r0 in a more presentation-friendly format.] Notice: This document has been prepared to assist the IEEE P802.15. It is offered as a basis for discussion and is not binding on the contributing individual(s) or organization(s). The material in this document is subject to change in form and content after further study. The contributor(s) reserve(s) the right to add, amend or withdraw material contained herein. Release: The contributor acknowledges and accepts that this contribution becomes the property of IEEE and may be made publicly available by P802.15. Razzell(Philips), Dabak(TI)

2. Questions and Answers on the properties of the MB-OFDM PHY Charles Razzell (Philips) Anand Dabak (TI) Razzell(Philips), Dabak(TI)

3. Won’t OFDM signals cause more interference than a noise-like signal? • The use of a scrambler for each sub-carrier ensures each 4MHz band has DSSS like spectral qualities • Merely frequency shifting these signals will not alter their spectrum or statistics • The IFFT can be seen as the summation of an efficient bank of frequency shifters • Net result is very flat spectrum random and Guassian amplitude distribution (CLT). Razzell(Philips), Dabak(TI)

4. Isn’t OFDM fundamentally the wrong choice in non bandwidth limited systems? • OFDM = Orthogonal Frequency Division Multiplexing • Choosing OFDM doesn’t dictate high order modulation schemes. • We chose QPSK with several additional forms of redundancy (FEC, repetition) • Historical associations (11a, DSL) should not cloud our engineering judgment. Razzell(Philips), Dabak(TI)

5. Doesn’t OFDM require much higher SNRs than binary DSSS modulations? • QPSK modulation combined with soft decision K=7 convolutional FEC is very robust. • For 110, 220 Mbps modulation/coding schemes MB-OFDM outperforms DS-CDMA • The required operating Eb/No for the MB-OFDM scheme is approximately 4dB for 8% PER on 1024Byte packets • The MBOK modulation schemes proposed by the DSSS proponents are not that simple! Razzell(Philips), Dabak(TI)

6. Doesn’t the MB-OFDM system require extremely high precision LO frequency matching? • Tolerable frequency offset is proportional to sub-carrier spacing • With QPSK, 5% of sub-carrier spacing is tolerable with low performance loss • 5% of 4.125MHz is 206.25Hz, which is >50ppm for Fc=4GHz. • This is much easier to accomplish than it was for 802.11a with its 312.5kHz sub-carrier spacing. Razzell(Philips), Dabak(TI)

7. Constellation Scatter Diagram with 206.25kHz LO frequency offset Razzell(Philips), Dabak(TI)

8. Don’t OFDM systems require very high linearity and I/Q matching for successful demodulation? • Again in danger of guilt by association with 802.11a • We use 4-QAM at all data rates: not 64-QAM. • This implies we can tolerate higher levels of intermodulation and I/Q distortions • The success of multiple, highly integrated 802.11a/g receivers that don’t enjoy this relaxation provides reason for confidence. Razzell(Philips), Dabak(TI)

9. Doesn’t the performance of OFDM systems suffer due to non-coherent processing of bandwidth? • Each of OFDM the sub-carriers potentially sees a different channel tap weight • The phase derotator after the FFT block at the receiver applies a complex conjugate channel tap weight to each received FFT bin. • The phase corrected and amplitude weighted symbols are then summed: • By additive combining of repeated symbols • In the Viterbi decoder during Add-Compare-Select • Both (1) and (2) can be seen as Maximal Ratio Combining of diversity components. Razzell(Philips), Dabak(TI)

10. Free distance, dfree tone 1 output tone 2 output tone 3 output tone i output tone i+1 output tone i+1 output tone i+3 output Illustration of Viterbi combining of Diversity Symbols Razzell(Philips), Dabak(TI)

11. The only way for OFDM systems to out perform DS-CDMA is with a back channel to stuff more bits on the tones with large SNR. • With a strong code as used in MB-OFDM, MB-OFDM performs coherent processing • Outperforms DS-CDMA • No back channel is needed to achieve this. • QPSK modulation is transmitted on all OFDM carriers of the MB-OFDM. • Back channel based techniques further improve the performance of OFDM systems where bit stuffing can be done on a per channel basis. • Back channel methods typically difficult for wireless channels due to bandwidth required • MB-OFDM does not need to employ any back channel for its improved performance over DS-CDMA. Razzell(Philips), Dabak(TI)

12. Given the coherent processing bandwith of MB-OFDM is 528 MHz, how can you claim MB-OFDM has comparable ranging capability to DS-CDMA? • Each band of the MB-OFDM occupies 528 MHz, the overall bandwidth of MB-OFDM mode 1 is 3*528 MHz = 1.584 GHz. • Coherent processing techniques can be employed for combining ranging data from the different bands giving an overall coherent processing bandwidth of 1.584 GHz. • MB-OFDM’s coherent processing bandwidth is the same as DS-CDMA Razzell(Philips), Dabak(TI)

13. Conclusions • The MB-OFDM signal is a very close approximation to AWGN, both statistically and spectrally • Use of OFDM for low bits/sec/Hz systems is both valid and advantageous with QPSK and multiple forms of redundancy • The choice of QPSK with powerful convolutional coding gives MB-OFDM robust performance Eb/N0=4dB • MB-OFDM does not require challenging frequency synchronization • Linearity and I/Q matching requirement are relaxed for the MB-OFDM PHY due to the use of low order modulation (QPSK) • MB-OFDM is well suited to (Maximal Ratio) coherent combining of multiple forms of redundancy • MB-OFDM does offer accurate ranging capability using the time-domain preamble part of the bursts. Razzell(Philips), Dabak(TI)