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:An overview of Signal Processing options for a new SUN PHY Date Submitted:March 1, 2009 Source:Steve Shearer, independent consultant 3655 Bernal Ave, Pleasanton, CA 94566 Voice: +1 925 417 1137, Cell: +1 925 997 0576 Re: [802.15.4g] TG4g Call for Preliminary Proposals, 2 February, 2009 Abstract:Some of the original requirements that drove 802.15.4 have changed substantially and a network that forms the basis of a Smart Grid must be highly reliable. This presentation provides a qualitative view on some aspects of modern signal processing that could be applied to these more complicated requirements. Characteristics of the new radio channels are discussed so that an appropriate PHY can be designed and tested. While the focus of the presentation is on maximizing spectral efficiency in order to achieve better reliability, the impact of complexity on power consumption and implementation cost are also addressed. Purpose: Proposal to add OFDM technology in addition to SSFH as part of one of the protocols defined in 802.15.4g 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. Steve Shearer, Independent Consultant

2. An overview of Signal Processing options for a new SUN PHY IEEE 802 Plenary Session Vancouver, BC March 2009 Steve Shearer. Independent Consultant

3. Original Requirements • Relatively short range • Low data rate applications • Very simple PHY and MAC • Home control, Industrial control • Simple control / monitoring type applications • Consequences of non-reliability are localised

4. Requirements are Changing • Longer range • Higher data rates • Larger packets • Bigger networks • More complex topologies • New applications like Smart Utility Networks • Mission Critical • Reliability is of utmost importance • consequences could be of National significance

5. Basic Components of Reliability • Accurate channel modeling is needed to design an appropriate PHY • Especially for physically longer links • Spatial diversity • Spectral efficiency affects link margin • Better modulation methods • More powerful channel coding • Improved PER affects ARQ performance • Better PER means fewer repeated packets • More efficient ARQ leads to a more reliable network

6. Channel models • AWGN is often appropriate for short ranges • BER has nice downward trend with SNR • Longer range channels exhibit fading • BER curve straightens and moves to the right • Longer channels can also be dispersive • multipath results in an irreducible error rate floor

7. Channel Models • The SUN application is unlikely to see rapid fading because the endpoints are generally static • But reflections can cause spatial nulls due to destructive interference • Some terminals may be “in a bad spot” • These spatial nulls can move with changes in the environment • New buildings, growing trees, etc.. • Repositioning equipment is expensive Steve Shearer, Independent Consultant

8. Suggestions for Mitigation • This WG should agree on a set of channel models for a number of deployment scenarios • Fundamental requirement to design and test a new PHY • Use demodulation methods that are robust against multipath and fading • Consider the use of simple spatial diversity techniques to mitigate spatial nulls Steve Shearer, Independent Consultant

9. SNR evaluation • Currently most evaluations of ZigBee plot Error Rate vs SNR • Using SNR makes it complicated to compare systems especially when channel coding is applied • because there is a difference between the information rate and the channel bandwidth • This figure indicates that 802.11b (1Mbps) performs better than 802.11b (11Mbps) on an SNR basis( this is correct, and is the reason why data rates are reduced to maintain connectivity in poor conditions) Steve Shearer, Independent Consultant

10. Spectral Efficiency • EbN0 is a normalised SNR measure and describes • Eb = energy per transported information bit (net of coding overhead) • N0 = noise spectral density • EbN0 analysis shows that WLAN at 11Mbps is more efficientthan WLAN at 1Mbps • Eb/N0 is a more consistent method for comparison of communication systems Steve Shearer, Independent Consultant

11. Suggestion • This WG should use EbN0 as a mechanism for comparison of new PHY proposals Steve Shearer, Independent Consultant

12. Modulation Methods • Very simple modulation systems like FSK are not spectrally efficient • Meaning that more energy is required to transmit the information bits, from source to destination, than necessary • More sophisticated modulation schemes achieve several dB’s better spectral efficiency • And often are able to produce optimal soft-decision information which helps the channel decoder Steve Shearer, Independent Consultant

13. Benefits of OFDM • OFDM has inherent frequency diversity without shortening the symbol time • Robust against multipath • Data rates are highly scalable without significant loss of efficiency • QAM on each carrier for high rates • Switch to BPSK on each carrier for lower rates • Use symbol repetition for even lower rates • Optimal soft decisions are easily obtainable to • Enhance the efficiency of the channel decoder • Mitigate frequency selective fading • Implement Maximal Ratio Combining (MRC) for diversity Steve Shearer, Independent Consultant

14. Suggestion • This WG should consider modulation methods used by proven state-of-the-art systems such as UWB, WLAN and WiMax Steve Shearer, Independent Consultant

15. Channel Coding • Modern channel coding provides spectral efficiency commensurate with its complexity • Turbo codes might be an overkill ! • But a simple concatenated coding scheme will offer significant benefits at modest cost Steve Shearer, Independent Consultant

16. Suggestion • This WG should consider state-of-the-art channel coding schemes and balance complexity vs required performance Steve Shearer, Independent Consultant

17. Implementation Costs • Characteristics of a single chip UWB PHY • Dual band 3GHz and 6Ghz operation • 1.5GHz operating bandwidth • Triple PLL’s to support 300ns , symbol by symbol frequency hopping • Complete analog / digital RX and TX chain including PA • Dual 7 bit 1Gsample/s ADCs • 128 pt Complex FFT, frequency domain equaliser • K=7 Viterbi decoder • Reed Solomon decoder • Low cost • Battery powered • Potential reqts for a ZigBee single chip PHY • Dual band 800 / 900 MHz operation • 2 MHz operating bandwidth • Single PLL, no frequency hopping • Complete analog / digital RX and TX chain including PA • Dual 7 bit 2M/sample/s ADCs • 32 pt Complex FFT, frequency domain equaliser • K=5 Viterbi decoder • Reed Solomon decoder • Low cost • Battery powered • Low power UWB PHY’s exist typically using 8 mm2 of silicon • We should be able to design a Zigbee PHY with state-of-the-art performance at substantially lower cost than that of UWB

18. Receive “active mode” consumption scales with modem complexity: The most significant consumers in UWB are the FFT and the Viterbi decoder UWB uses a K=7 Viterbi decoder, a ZigBee product might use a K=5 decoder ~ 1 / 2(7-5) = ¼ the power consumption UWB uses a 128 pt FFT, a ZigBee product might use a 32 pt FFT ~ 1 / 2(2) = ¼ the power consumption Receive “active mode” consumption also scales with output data rate UWB = 480Mbps, ZigBee might be 1Mbps Overall ZigBee power consumption could be as low as 1/2000 that of UWB Power Consumption Expectations Steve Shearer, Independent Consultant

19. Conclusion • Smart Utility Networks represent a significant change in the requirements of an 802.15.4 network • One key requirement is reliability • reliability is linked to spectral efficiency and accurate channel modeling • Modern signal processing approaches provide better spectral efficiency • We should be using proven state-of-the art methods scaled to the requirements of SUN systems