Ultra-Wide Band Communication for the Internet of Things

Ultra-Wide Band Communication for the Internet of Things The MICS UWB Network uwb. epfl. ch Jean-Yves Le Boudec (coordinator), EPFL I&C 21 -23 January 2008 1

Abstract: Ultra-Wide Band communication is a technology for low range, low power sensor and mobile devices which employs very low transmission powers (below the level of unintentional emissions) and high bandwidth. It possesses a number of unique features that make it very attractive to many local applications. First, ranging with high accuracy is possible even indoors. Second, it is resistant to multipath fading which often pleagues indoors communications. Third, it scales well in dense deployments. Fourth, cryptographic modulation is possible. In this talk, we describe the research done in the MICS Ultra-Wide Band network, showing ranging, dense deployment capabilities and medical applications. 2

Table of Contents 1. The UWB Network of MICS 2. What is UWB ? 3. Impulse Radio UWB 4. Low Power Medical Application 5. Robustness to Interference 6. Ranging 7. Outlook 3

A Network within MICS researching on Impulse Radio UWB The network CSEM, Neuchatel Prof. Farserotu, Hai Zhan Prof. Decotignie, Jerôme Rousselot ETHZ, Zurich Prof. Wittneben, Florian Trösch, Christoph Steiner EPFL I&C, Lausanne Prof Le Boudec (coordinator), Ruben Merz, Manuel Flury Impulse radio Ultra Wide Band communication Low power In presence of multi user interference Ranging Provide fundamental research and proofs of concept EPFL STI, Lausanne Prof. Dehollain, James Colli. Vignarelli, Prakash Thoppayegambaram Prof. Skrivervik, Gabriela Quintero HES SO, Yverdon Prof. Robert, Jérome Vernez ST Microelectronics, Geneva Dr. J. Zory 4

Table of Contents 1. The UWB Network of MICS 2. What is UWB ? 3. Impulse Radio UWB 4. Low Power Medical Application 5. Robustness to Interference 6. Ranging 7. Outlook 5

Ultra Wide Band (UWB) Communication Use a very large spectrum up to Several GHzs Very low power Below level of unintentional emission Power Limits FCC (2002) limits peak power (0 d. Bm per 50 MHz) mean power (-41. 3 d. Bm per MHz) Europe (and CH-Ofcom, 2007) put more stringent limits Unlicensed Co-exists with other technologies US EC (source: FCC 2002, CH-Ofcom, 2007) 6

Various Uses of UWB Signals Radar and Ranging Radar A very old UWB application, used for maritime or air navigation, and as remote speedometer New apps: automotive security, rescue operation One active device analyzes echo Target is passive and unaware of signal Not always low power E. L. Ranging From device to device Device is active sender Base station is receiver /transmitter E. g Ubisense, Cambridge UK Low power E. L. 7

Various Uses of UWB Signals Communication Short Range Communication Low power Up to 30 m indoors High data rate UWB Communication Wireless USB / Wireless Firewire Uses entire bandwidth Very large bit rate on one single link Peaky in frequency Low data rate E. g. Sensor networks Impulse radio signals Very large aggregate throughput Robots with ranging needs for collective intelligence Source: Prof. Alcherio Martinoli 8

Strengths and Weaknesses of UWB High throughput for high data rate Shannon-Hartley law: C = B log 2 ( 1 + S/N ) with C = bit rate (b/s) B = bandwidth (Hz) Exploited by Wireless USB / Firewire : 100 - 480 Mb/s for Wireless USB over 3 -10 m Low Power for Low Data rate Scalability Sensor network with very large bandwidth, total capacity scales with number of nodes Source: Mohammad Abualreesh Resistance to Channel Impairments Multiple paths are distinguishable Suitable for indoors, terrain with obstacles, metallic environment High Resolution in time domain Ranging with cm accuracy indoors Secure ranging Short range 10 m to 30 m 9

Table of Contents 1. The UWB Network of MICS 2. What is UWB ? 3. Impulse Radio UWB 4. Low Power Medical Application 5. Robustness to Interference 6. Ranging 7. Outlook 10

Impulse Radio UWB Uses Short Pulses are narrow in time, wide in frequency Pulse duration order of 1 ns Source: Gabriela Quintero Features Low power Duty cycle at 1 Mb/s = 1 % Robust against multi-user interference High precision ranging 11

Impulse Radio UWB Uses Time Hopping Sequence: […, 2, 5, 4, 7 …] Pulses appear random unless you know THS is predictible to user who knows the key ; e. g. : MAC address Transforms packet collision into symbol collision Increaed bit error rate instead of packet loss Software-like flexibility in hardware When a pulse is sent can easily be changed by modifying a few values in the system Change the time hopping sequence Change the modulation rate 12

Multipath Propagation Signal propagation subject to reflections Pulses are attenuated / modified but still distinguishable Very little destructive interference Channel response Received signal 13

Table of Contents 1. The UWB Network of MICS 2. What is UWB ? 3. Impulse Radio UWB 4. Low Power: Medical Application 5. Robustness to Interference 6. Ranging 7. Outlook 14

Body Area Network with UWB Requires very low power Very bad transmission channel UWB body area network prototype developed at ETH / Prof A. Wittneben’s group Ear to ear communication Focus on low power and point to point link 15

Wireless BAN Communication for less than 1 m. W Bursts of 500 bits/ms Average Data Rate of 500 kbits/s Peak Data Rate of 50 Mbits/s Simple Tx and Rx Structures Mainly Analog Processing Estimated Power Consumption < 1 m. W Analog Part Low Cost Low Power Low Complexity Ultra-Wideband Radio Rx Chain Energy Detection Tx Chain UWB Pulse Generator 1% duty cycle 500 kbits/s < 0. 3 m. W Sampling at 200 MHz Digital Baseband ADC Clock Synthesis Synchronization Decoding Error Correction MAC < 0. 7 m. W 16

Body Area Network UWB Test Bed Ear-to-Ear Channel 17

GUI for UWB test-bed Average transmit power -45 d. Bm Ear-to-ear channel with artificial waterbucket-head BER at -45 d. Bm is 0. 04, capacity is 480 Mb/s transmit receive 18

Relevant Publications F. Troesch, C. Steiner, T. Zasowski, T. Burger, and A. Wittneben, "Hardware Aware Optimization of an Ultra Low Power UWB Communication System, " IEEE International Conference on Ultra-Wideband, ICUWB 2007, Marina Mandarin, Singapore, Sept. 2007. C. Steiner and A. Wittneben, "On the Interference Robustness of Ultra. Wideband Energy Detection Receivers, " IEEE International Conference on Ultra -Wideband, ICUWB 2007, Singapore, Sept. 2007. 19

Table of Contents 1. The UWB Network of MICS 2. What is UWB ? 3. Impulse Radio UWB 4. Low Power Medical Application 5. Robustness to Interference 6. Ranging 7. Outlook 20

Robustness to Interference From Theory to Practice In Theory, UWB transmission is robust to interference from other UWB systems Due to large bandwidth In practice, this requires careful system design MAC Signal Acquisition Accommodate multipath This makes UWB systems potentiallyscalable, well adapted to dense deployments Throughput per node constant with number of nodes N Contrast to narrowband systems: » N-1/2 21

PHY-Aware MAC Classical organization of a network E. g. Wi. Fi, Bluetooth PHY transmits packets MAC avoids collisions i. e. MAC = mutual exclusion This is not efficient for UWB Mutual exclusion divides throughput linearly… … but most collisions are at pulse level Rate reduction is small The optimal is: Allow interference and manage it ! Requires MAC to be PHY aware Our experimental MAC B A C Data THS(A), Code = Ri THS(A), Code = Rj NACK THS(A), Code = RN Incremental Red. THS(A) ACK Interference, not collision THS(A), Code = RN Idle THS(B), Code = RN 22

DCC-MAC A PHY aware MAC protocol, designed to be robust to interference DCC= dynamic channel coding Key features of design N nodes in a chain One time hopping sequence per destination (private time hopping sequences) Interference mitigation at pulse level Mutual exclusion for a single destination only Rate adaptation DCC MAC CA/CDMA -like 802. 11 like 23

Signal Acquisition Signal acquisition is difficult for Impulse Radio UWB Signal is intermittent Interferences are allowed Classical methods based on gaussian noise hypotheses do not apply Power Independent Detection (PID) is robust to interference even if interfering power is larger than intended signal uses thresholding 24

Private Time Hopping Sequences Common Time Hopping Sequence in preamble Many useless acquisitions One Private Time Hopping Sequence per destination Acquisition is private, only intended receiver decodes Requires source to know sequence of destination E. g. linear congruence seeded with MAC address of destination 25

Private Sequences Avoid the Ad-Hoc Collapse Ad-hoc collapse Many TCP connections in an ad-hoc Collapses with 802. 11 and other protocols Due to collisions No good solution known to this problem With private sequences, the ad -hoc collapse goes away Nodes acquire only packets destined to self 26

Accommodate Multipath Assume modulation is pulse position With interferers and multipath, received signal looks like 27

Idea: (Rake receiver) Estimate channel during signal acquisition phase Look for pattern of pulses in the received signal - correlation Use thresholds to avoid near end effects Similar ideas apply to energy detectors 0 28

Performance Evaluation of IEEE 802. 15. 4 a Standard for Impulse Radio UWB, Low Data Rate MAC influenced by narrow band tradition 2 THSs in total Makes some compromises to ease implementation Bursts of pulses Q: how does it perform with respect to interference robustness ? Multiple transmissions in same network Transmissions from neighbouring, non coordinated network We simulated the standard in detail, with interferers, and compared its performance against two benchmarks Benchmark 1: Destructive collision Packet lost when two transmissions overlap ALOHA performance Typical of narrowband systems Benchmark 2: Perfect capture Packets compete during signal acquisition and transmission Only one succeeds Typical of ideal UWB system 29

IEEE 802. 15. 4 a is not Robust to Interference Benchmark 1: Destructive collision 802. 154 a, with interference Benchmark 2: Perfect capture 802. 154 a, no interference Performance is close to destructive collision Does not exploit UWB benefits well Possible fixes Compress bursts Private time hopping sequences 30

Interference Testbed Goal: Implement and test multi-user impulse radio system In presence of multi-user interference Real hardware, still programmable in matlab A coordinated effort of the MICS UWB network Ruben Merz (coordinator) James Colli-Vignarelli Gabriela Quintero Prakash Thoppayegambaram Jerome Vernez Jean-François Zürcher 31

Interference Testbed (EPFL, HES SO) Video by Jerome Vernez, HES SO (Yverdon) 32

Relevant Publications El Fawal, Alaeddine ; Le Boudec, Jean-Yves, “A Robust Signal Detection Method for Ultra Wide Band (UWB) Networks with Uncontrolled Interference”, In: IEEE Transactions on Microwave Theory and Techniques (MTT), vol. 54, num. 4, part 2, 2006, p. 1769 -1781 Radunovic, Bozidar ; Le Boudec, Jean-Yves, “Optimal Power Control, Scheduling and Routing in UWB Networks”, In: IEEE Journal on Selected Areas in Communications, vol. 22, num. 7, 2004, p. 1252 Merz, Ruben ; Widmer, Jörg ; Le Boudec, Jean-Yves ; Radunovic, Bozidar, “A Joint PHY/MAC Architecture for Low-Radiated Power TH-UWB Wireless Ad-Hoc Networks”, In: Wireless Communications and Mobile Computing Journal, Special Issue on Ultrawideband (UWB) Communications, vol. 5, num. 5, 2005, p. 567 -580 Flury, Manuel ; Merz, Ruben ; Le Boudec, Jean-Yves, “Managing Impulsive Interference in Impulse Radio UWB Networks”, In: ST Journal of Research, 2007 Flury, Manuel ; Merz, Ruben ; Le Boudec, Jean-Yves ; Zory, Julien, “Performance Evaluation of an IEEE 802. 15. 4 a Physical Layer with Energy Detection and Multi-User Interference”, In: IEEE International Conference on Ultra-Wideband (ICUWB 2007), 2007 33

Table of Contents 1. The UWB Network of MICS 2. What is UWB ? 3. Impulse Radio UWB 4. Low Power Medical Application 5. Robustness to Interference 6. Ranging 7. Outlook 34

Impulse Radio UWB enables low cost ranging at high precision Short pulses can easily be located by receiver Basis for radars Can be used at low cost in all sorts of equipments with UWB 2 techniques are researched in the MICS UWB Network Geo-regioning High resolution ranging 35

Geo-Regioning Channel response A method for location fingerprinting Idea: channel impulse response is correlated in space Method: Learning phase: send test signals to base station from various locations Analyze correlations (e. g. covariance matrix, delay profile) Tracking Phase Mobile sends beacons to base station Real time correlation is performed 36

UWB Geo-Regioning Demonstration Developed by Prof. A. Wittneben’s group / ETHZ Channel impulse responses from region 22 to RX 37

Relevant Publications C. Steiner, F. Althaus, F. Troesch, and A. Wittneben, "Ultra-Wideband Geo. Regioning: A Novel Clustering and Localization Technique, " EURASIP Journal on Advances in Signal Processing, Special Issue on Signal Processing for Location Estimation and Tracking in Wireless Environments, Nov. 2007. C. Steiner and A. Wittneben, "Clustering of Wireless Sensors based on Ultra. Wideband Geo-Regioning, " Asilomar Conference on Signals, Systems, and Computers, Pacific Grove, USA, Nov. 2007. 38

High Resolution Ranging Accurate ranging = estimation of distance Based on time of arrival of signal Idea: mobile sends UWB pulses to one or several base stations detect first pulse at receiver How: Estimate both channel response and time of arrival of first pulse Not always strongest Remove noise and interference by modified Prony algo 39

Ranging Through Obstacles and With Interferers Experimental setting Click on figure for video Non severe non light of sight ranging is possible E. g. through wood or cardboard The modified Prony algorithm finds the first pulse Sent signal contains a train of encoded pulses Received signal contains many replicas due to multipath Strong pulses help find weak Video by Hai Zhan, CSEM Quiet room at EPFL (not anechoic) Experiment implemented by Hai Zhan (CSME) True distance is 48. 8 cm – estimated distance is 50. 0 cm 40

Relevant Publications Zhan, Hai ; Farserotu, John ; Le Boudec, Jean-Yves “A Novel Maximum Likelihood Estimation Of Superimposed Exponential Signals In Noise And Ultra-Wideband”, PIMRC 07, 2007 Zhan, Hai ; Ayadi, Jaouhar ; Farserotu, John ; Le Boudec, Jean. Yves, “High-Resolution Impulse Radio Ultra Wideband”, In: The 2007 IEEE International Conference on Ultra-Wideband, ICUWB 2007, 2007 41

Table of Contents 1. The UWB Network of MICS 2. What is UWB ? 3. Impulse Radio UWB 4. Low Power Medical Application 5. Ranging 6. Robustness to Interference 7. Outlook 42

Impulse Radio UWB is a key technology for the Internet of Things Unique features Indoors ranging Resistance to multiuser interference Scalable total throughput Very low power Potential areas of future research Secure ranging Very short signal time High throughput ranging Frequent position updates for distributed robot control Practical developments are only starting Standard based implementations can be improved 43

Thank You Special thanks go to all who helped prepare this presentation Jerome Vernez Hai Zhan Ruben Merz Christoph Steiner And to all other contributors of the MICS UWB network who make this project such a great fun Manuel Flury James Colli-Vignarelli Jean-Dominique Decotignie Catherine Dehollain John Farserotu Gabriela Quintero Stephan Robert Jérome Rousselot Anja Skrivervik Prakash Thoppayegambaram Florian Trösch Armin Wittneben Julien Zory 44