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 Evaluation of Indoor Ultra-Wideband Communication Channels] Date Submitted: [8July, 2002] Source: [R. Jean-Marc Cramer] Company [TRW Space & Electronics] Company [] Address [One Space Park DriveMail Stop 02/2743Redondo Beach, CA 90278 ] Voice [310-812-9073], FAX: [] E-Mail:[jean-marc.cramer@trw.com] Re:[] Abstract: [] Purpose: [UWB channel model presentation.] 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. J.M. Cramer, TRW Space and Electronics

2. An Evaluation of Indoor Ultra-Wideband Communication Channels R. Jean-Marc Cramer jean-marc.cramer@trw.com 9 July, 2002 J.M. Cramer, TRW Space and Electronics

3. Overview • Introduction and motivation • UWB propagation experiment • UWB array signal processing • Application to measured data • Channel models for UWB signal propagation • Conclusions J.M. Cramer, TRW Space and Electronics

4. Introduction and Motivation • Accurate models of the propagation channel are required for the design of UWB systems • UWB radio algorithm verification and design trades • Performance prediction • Previous studies have reported characterizations of more narrowband channels • These studies may not adequately reflect the bandwidth-dependent effects of the propagation of UWB signals. • Propose channel models of the form: where the pulse shape and amplitude are dependent on the particular multipath component J.M. Cramer, TRW Space and Electronics

5. UWB Propagation Experiment • Measurements made at 14 different locations in an office building • At each location data collected on a 7x7 array of sensors with 6” spacing • Array located 1.65 meters above the floor and 1.05 meters below the ceiling Pulse at 1m separation from the transmit antenna (Direct path pulse) J.M. Cramer, TRW Space and Electronics

6. Typical Received Signal Profiles Measurement array J.M. Cramer, TRW Space and Electronics

7. Sensor-CLEAN Algorithm • Processed the measured data with a variation of the CLEAN algorithm • Delay-and-sum beamforming used as a first step • Constructed beamformer response over of direction and time • Signal detection at peak in the response over time and angle • Only assumption on the incident signal is that it exists in a short window of time at beamformer output • No canonical wave shape is assumed • Permits study of incident waveform shape and signal statistics • Iteratively reduce sensor data based on detected signal locations (peaks in the beamformer output) • Generate updated beamformer output and repeat until residual threshold is satisfied • Refer to the algorithm used here as Sensor-CLEAN • Relaxation step conducted directly on the sensor data J.M. Cramer, TRW Space and Electronics

8. UWB Channel Modeling • UWB signal processing techniques applied to measured propagation data • Resulting recovered UWB signal information used to study • Received waveform shapes • Path-loss models • Clustering models for indoor signal propagation • Spatio-temporal distributions of the received signal energy • Ray tracing based on single bounce elliptical models • UWB channel synthesis J.M. Cramer, TRW Space and Electronics

9. Application of Algorithm to Measured Data • Recover measurement locations from time-of-arrival and azimuth angle-of-arrival of first detected signal • Corresponding direct path length is • X indicates recovered measurement locations • Squares are actual measurement locations J.M. Cramer, TRW Space and Electronics

10. Recovered Signals at Location P • Plot shows recovered signal amplitude versus time and azimuth at P • Dependence on elevation angle has been integrated out • Existence of clusters of multipath signals is noted is the plots LOS signal at 5.47m and 49o J.M. Cramer, TRW Space and Electronics

11. Recovered Signals at Location H LOS signal at 10.2m and 149o J.M. Cramer, TRW Space and Electronics

12. Recovered Signals at Location M LOS signal at 13.07m and 255o J.M. Cramer, TRW Space and Electronics

13. dLOS = 5.47 m Location P Recovered LOS Waveforms • Each plots displays two curves corresponding to the waveform recovered by processing algorithms using different time windows • A larger window gives more complete picture of isolated signals • Smaller window is better for resolving dense multipath • Compare to direct path waveform shown earlier Location F2 dLOS = 5.61 m J.M. Cramer, TRW Space and Electronics

14. Clustering Models for UWB Propagation • Previous models for indoor propagation reported the existence of clusters of multipath components • This model is justified for the UWB channel by the apparent existence of clusters seen in the scatter plots • Further justified by use of a sliding window on the recovered signal information over time and angle • Assume channel impulse response is separable as a function of time and azimuth angle: • Final cluster locations determined by manual inspection of the recovered signal locations over time and angle • 65 clusters identified at 14 different measurement locations J.M. Cramer, TRW Space and Electronics

15. Clustering Models for UWB Propagation Sliding window over the recovered signals at location M shows the existence of clusters of multipath J.M. Cramer, TRW Space and Electronics

16. Clustering Models for UWB Propagation • Model the received signal amplitude by a Rayleigh distributed RV with MS value given by, • is the average power in first arrival of first cluster • The variable Tl represents the arrival time of the lth cluster • The variable tkl is the arrival time of the kth signal within the lth cluster, relative to Tl • The parameters G and g determine the inter-cluster (cluster) and intra-cluster (or ray) rates of decay • The exponent G is generally determined by building architecture • The exponent g is determined by objects close to the receive antenna J.M. Cramer, TRW Space and Electronics

17. Inter-Cluster Decay Rate J.M. Cramer, TRW Space and Electronics

18. Intra-Cluster Decay Rate J.M. Cramer, TRW Space and Electronics

19. Energy Deviation from Mean • Deviation from the mean is hypothesized to follow Rayleigh dist • This distribution represents best-fit to the recovered UWB signal information when considering Rayleigh, lognormal, Nakagami-m and Rician distributions J.M. Cramer, TRW Space and Electronics

20. Cluster Angle-of-Arrival • Cluster angle-of-arrival is hypothesized to follow a uniform distribution • Ray angle-of-arrival is hypothesized to follow a Laplacian distribution, according to • Recovered ray or intra-cluster arrivals were tested against Gaussian and Laplacian hypothesis. • Best fit distribution and resulting standard deviation were reported. • Laplacian with s=38° is the best fit distribution. J.M. Cramer, TRW Space and Electronics

21. Cluster and Ray Angles-of-Arrival s=38° • Distribution of cluster angles is relatively uniform below 135o • Additional measurement would probably provide clusters above 135o J.M. Cramer, TRW Space and Electronics

22. Rate of Signal Arrivals • Signal inter-arrival times are hypothesized to follow an exponential rate law according to, where L is the cluster arrival rate and l is the ray arrival rate • Following this model, best-fit distributions were determined for recovered UWB signal arrival times • Faster arrival rates were found than in previous reports • Could be due at least in part to finer time resolution J.M. Cramer, TRW Space and Electronics

23. Ray Arrival Rates J.M. Cramer, TRW Space and Electronics

24. UWB Channel Parameter Summary • Many potential reasons for differences in the values. • Finer time resolution of UWB signals. • Larger fractional bandwidth and better penetration performance of UWB signals. • Differences in the building materials and/or architecture. • Differences in the orientation of the transmitter and the receiver • Bottom line: More UWB studies need to be made before concrete conclusions can be drawn J.M. Cramer, TRW Space and Electronics

25. UWB Channel Synthesis • Synthesis of UWB channels permits accurate simulations of UWB systems • Multipath arrivals for two UWB channels synthesized from the clustering model parameters are shown below J.M. Cramer, TRW Space and Electronics

26. Conclusions • UWB signal processing techniques applied to data measured on an array of sensors • Resulting received signal information used to develop models of the indoor UWB propagation channel • Comparisons drawn with more narrowband channel models • UWB channel parameters derived from measurements taken in a single building with a single type of UWB pulse • Building architecture and the geometry of the experiment can impact the received signal statistics • Statistics of other UWB waveforms may be different • Main result may be development of techniques for analysis and processing of UWB signals • Techniques can be applied to other UWB waveforms J.M. Cramer, TRW Space and Electronics