Implementation challenges of UWB for sensor networks Laurent

Implementation challenges of UWB for sensor networks Laurent Chalard, Didier Hélal, Gian-Mario Maggio, Yinqwei Qiu, Lucille Verbaere-Rouault, Armin Wellig, Julien Zory STMicroelectronics, Geneva, Switzerland UWB 4 SN – November 2005

Content • • • Introduction Market/Application requirements Regulation Standardization Wireless Sensor Mote – – Link budget MAC Synchronization FEC • Conclusions UWB 4 SN – November 2005

A problem under constraints Market understanding Complete mote solution Competitive advantage Innovative WSN solutions Standard compliancy ST’s technology compatibility Ready market UWB 4 SN – for November 2005

Major Limitations to Global Wireless Sensor Adoption Ease/install Reliability Interference Battery Cost Interoperability Security Bit rate No need Size Source ON-World 2004 UWB 4 SN – November 2005

Application requirements expressed at IEEE 802. 15. 4 a Applications' requirements ta a d Low es not do ! e e l t p a r im s n mea Data rate (kbps) 1000 10 1 1 10 1000 Range (meter) + ranging with accuracy inside 5% of range UWB 4 SN – November 2005 10000

Regulation • No harmonization done by ITU-R so far… • EC final decision in March 2006 • Low data rate is still under discussions – duty cycle • Minimum average burst repetition period over an hour 1 sec • Minimum instantaneous burst repetition period over 1 second 30 to 200 ms – emission level limitation -41. 3 d. Bm/MHz or -45 d. Bm/MHz UWB 4 SN – November 2005

Standardization (1) • Standards has exhibited limitations up to know for wireless sensor network applications • • • 802. 15. 4: Zigbee: Wi. Fi: BT: poor reliability too complex too expensive limited in number of nodes Now appear 3 different alternate PHY options in IEEE 802. 15. 4 a • • • Low-band UWB [DC-960 MHz] Chirp Spread Spectrum [2. 4 GHz ISM band] High band UWB [3. 1 -10. 6 GHz] UWB 4 SN – November 2005

Standardization (2) IEEE 802. 15. 4 a status • Band plan defined • PRF will be a multiple of 7. 21875 MHz • Perfect Balanced Ternary Sequences (PBTS) of length 31 and 127 have been agreed. • All systems should support a mandatory non-coherent mode • Still 6 Forward Error Correction proposals (Super Orthogonal Codes, Convolutional codes) UWB 4 SN – November 2005

Mote: Complete Solution • Fully integrated wireless sensor devices – – Small: < 1 cm 3 (System-in-Package ) Cheap: <1$ (low cost electronics) Low power: <10 m. W peak Operate from energy scavenging: <100 u. W average Energy scavenging Sensor Actuator A D C D I O D A C Battery Capacitors Power management Calibration -controller BB + RF transceiver TEDS UWB 4 SN – November 2005

STMicroelectronics – AST areas of work in WSN • 802. 15. 4 / Zig. Bee (PHY, MAC and networking protocol) • UWB Physical Layer • Localization enabled networking Target is convergence ! UWB 4 SN – November 2005

Mote: MAC • • Support for ranging procedures (including mobility) Backward compatibility (w. r. t. 802. 15. 4 MAC) Cross-layer (PHY-MAC) optimization Medium access: – “Carrier Sensing” type mechanisms for UWB (to enable CSMA) – Random access schemes (e. g. ALOHA) • Interference mitigation: – LDC operation – DAA (Detect and Avoid) UWB 4 SN – November 2005

Example of a LDR budget link Regulation TX Power Path Loss RX Power Link Margin Implementation Loss System Noise per bit Eb/No min Noise Figure Data throughput Thermal Noise Temperature UWB 4 SN – November 2005

Synchronization (1) • Context – Inaccurate reference clocks (typ. >> 1 ppm) – Multi-user, asynchronous & random communications – Low SNR => Need for pulse energy accumulation (CI and/or MF, etc. ) – Short pulses => down-convert to limit processing speed • Synchronization shall overcome… – – – – Jitter (reference clock & PLL) Drift between motes’ clocks (frequency offset) Noise Interferences multi-user Mobility etc. UWB 4 SN – November 2005

Synchronization (2) • A few illustrative numbers… – Coherency time of a 500 MHz pulse is in the order of 100 ps – Preamble duration is between 1 us and 33 us – Possible drift due to oscillator's accuracy • Over 1 us, 10 ppm to ± 10 ps, 40 ppm to ± 40 ps • Over 33 us, 10 ppm to ± 330 ps, 40 ppm to ± 1. 32 ns • Hence a few design challenges – Acquisition/detection • How to coherently accumulate energy? • How to estimate frequency drift, so as to relax tracking requirements? – Tracking • How to do it on a non-continuous signal? • How to do it with minimum complexity? UWB 4 SN – November 2005

Super Orthogonal vs convolutional codes • Gap between SOC and CC decreases in dense multipath environment. • SOC performances for non coherent Rx ? UWB 4 SN – November 2005

FEC general requirements • Unique solution for coherent AND non coherent mode: puncturing ? K Coherent Rx AWGN Eb. N 0 @ 1%PER Non-coherent Rx AWGN Eb. N 0 @1%PER Uncoded 9. 0 d. B 12. 25 d. B 3 5. 2 d. B 9. 8 d. B 4 4. 5 d. B 9. 2 d. B 5 4. 0 d. B 8. 8 d. B • Trade-off Complexity vs performance – L constraint length – Nb operation per bit=2 L – hard decoding for min complexity (soft decoding -> ~2 d. B improvment) • (De)Interleaver mandatory to obtain good decoding performances. Eb/No requirements with convolutional codes, coding rate=1/2. UWB 4 SN – November 2005

RANGING To TOA & Two-Way Ranging T 1 Terminal A TX/RX Terminal B RX/TX TOF Terminal A Prescribed Protocol Request Delay and/or Processing Terminal Time TOF TReply TOF Estimation B TOA error + clock drift @ B Ranging error < 1 m TOAerror < 3. 3 ns Courtesy: LETI TOA error + clock drift @ A UWB 4 SN – November 2005

Symmetric Double Sided-Two Way Ranging (SDS -TWR) Device A Device B TOF Time of flight TRound. A TReply. B >> TOF reply time BUT: The condition TReplay. A TReplay. B limits the MAC protocol ! TOF TRound. B TReply. B TReply. A TReply. B TOF (IEEE 802. 15. 4 a: Nanotron) Two big numbers measured with the same time-base (clock B) Two big numbers measured with the same time-base (clock A) UWB 4 SN – November 2005

TOA estimation error • LOS • NLOS • Delay spread + AWGN + Relative clock drift between Terminal A and B matched filter output (of coherent bins) 80 ppm UWB 4 SN – November 2005

Conclusion • Only a complete wireless sensor network solution will enable the emergence of a mass market • This can only be achieved through cross optimization Market understanding Complete mote solution Competitive advantage Innovative WSN solutions Standard compliancy ST’s technology compatibility UWB 4 SN – November 2005