1. <month year>
doc.: IEEE /453r0-SG3a
November 2002
Project: IEEE P Working Group for Wireless Personal Area Networks (WPANs)
Submission Title: [Analysis of Time Variance of a UWB Propagation Channel (Slides)]
Date Submitted: [5 November 2002]
Source: [Steve Schell] Company [Bitzmo, Inc]
Address [2130 Gold St. Suite 250, P.O. Box 1180, Alviso CA 95002]
Voice:[ ], FAX: [ ],
Re: []
Abstract: [An analysis of Intel’s channel measurements suggests that a 500-MHz wide channel can change as little as dB (10% NRMSE) over a duration of 60 milliseconds in the presence of an object moving at pedestrian speeds near the line-of-sight path. The analysis also suggests that the channel can change completely (100% NRMSE) over the same or smaller duration. The corresponding assumptions on packet-to-packet correlation are “high correlation” and no correlation, respectively. A recommendation is made to consider both scenarios in the evaluation of PHY proposals.]
Purpose: [A sequence of UWB propagation channel impulse responses measured by Intel is analyzed to assist the a Study Group in determining whether to consider only propagation channels that are uncorrelated from packet to packet, or only those that are invariant from packet to packet, or both.]
Notice: This document has been prepared to assist the IEEE P 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 P
Steve Schell, Bitzmo Inc
<author>, <company>

2. Analysis of Time Variance of a UWB Propagation Channel*
November 2002
Analysis of Time Variance of a UWB Propagation Channel*
Steve Schell
Bitzmo, Inc
*see also 02/452 for accompanying paper
Acknowledgement: Thanks to David Cheung and colleagues at Intel for making their channel measurements (on which this work is based) available to the UWB community.
Steve Schell, Bitzmo Inc

3. Is a UWB channel correlated or uncorrelated from packet to packet?
<month year>
doc.: IEEE /453r0-SG3a
November 2002
Is a UWB channel correlated or uncorrelated from packet to packet?
Motivations:
understand the application environments in which a UWB-enabled consumer electronics device would operate
contribute to SG3a / TG3a PHY evaluation process
proposals that depend on time-variance or time-invariance should have weaknesses evaluated
proposals that can smartly exploit time-variance or time-invariance should be allowed to shine
Steve Schell, Bitzmo Inc
<author>, <company>

4. Starting Point: Intel measurement data
November 2002
Starting Point: Intel measurement data
Sequence of impulse responses versus position of a water container
2-8 GHz measurement band
data on USC ftp site
Steve Schell, Bitzmo Inc

5. Intuition: Channel correlation vs. position varies from high to low
November 2002
Intuition: Channel correlation vs. position varies from high to low
high
correlation
LOS: water blockage out on the edges
low
correlation
NLOS: water blockage
in the LOS path
Steve Schell, Bitzmo Inc

6. Analysis: Channel similarity from position to position
November 2002
Analysis: Channel similarity from position to position
Methodology
Compute prediction error between channel impulse responses x(t,f) & y(t,f) from adjacent positions, as a function of center frequency f :
prediction error
Steve Schell, Bitzmo Inc

7. Results: Channel similarity from position to position
November 2002
Results: Channel similarity from position to position
Results for 500 MHz bandwidth
{mean,min,max,etc} over all f of prediction_error(position)
Steve Schell, Bitzmo Inc

8. Results: Channel similarity from position to position
November 2002
Results: Channel similarity from position to position
Results for 2000 MHz bandwidth
{mean,min,max,etc} over all f of prediction_error(position)
Steve Schell, Bitzmo Inc

9. Discussion of position-by-position results
<month year>
doc.: IEEE /453r0-SG3a
November 2002
Discussion of position-by-position results
Transition regions
water moves from non-blocking to blocking LOS path
channel impulse response changes a lot
prediction error up to 0 dB NRMSE or 100 %
Stable regions
propagation remains LOS (or NLOS)
channel impulse response changes a little
prediction error as little as -20 dB NRMSE or 10 %
Overall
channel impulse response changes by anywhere from -20 dB (10% NRMSE) to 0 dB (100% NRMSE) with a 2-inch change in water position
Steve Schell, Bitzmo Inc
<author>, <company>

10. <month year>
doc.: IEEE /453r0-SG3a
November 2002
Re-interpretation of position-by-position results in terms of time variance
Example:
assume water blockage similar to pedestrian blockage
2 inch step at 3 km/hr  61 ms per step
1024-octet packet at 110 Mbps  74 μs per packet
≤ 800 packets per 2 inch step at pedestrian speeds
channel impulse response changes by anywhere from -20 dB (10% NRMSE) to 0 dB (100% NRMSE) over a duration ≤ 800 packets:
as little as 0.013% NRMSE change per packet
Steve Schell, Bitzmo Inc
<author>, <company>

11. Other discussion of results
<month year>
doc.: IEEE /453r0-SG3a
November 2002
Other discussion of results
Doppler spectrum
cannot be estimated for this data set
transition regions violate wide-sense-stationarity requirement
stable WSS contain too little data
Bandwidth dependencies
when blockage is moving, results for 500 MHz and 2000 MHz channel bandwidths are similar
situation changes when Tx or Rx is moving
time-variant propagation distance causes prediction error to increase with increasing bandwidth
simulated relative Tx-Rx motion at pedestrian speeds:
results for 500 MHz bandwidth very similar to stationary case
results for 2000 MHz bandwidth much worse (min error = 25%)
Center-frequency dependencies
prediction error (% NRMSE) increases mildly with f
mean 25% at 3 GHz, increases to 40% at 8 GHz
Steve Schell, Bitzmo Inc
<author>, <company>

12. Recommendation PHY proposal evaluation include 2 scenarios
<month year>
doc.: IEEE /453r0-SG3a
November 2002
Recommendation
PHY proposal evaluation include 2 scenarios
perfect packet-to-packet correlation
channel impulse response constant across 100’s of packets
example application scenario
highest data-rate, short-range data transfers
zero packet-to-packet correlation
channels for consecutive packets are independent
example application scenarios
wireless HDTV / home theater with scampering kids run amok
lower throughput connections (low packet rate)
Steve Schell, Bitzmo Inc
<author>, <company>