1. Submission Title: [Spectral Properties of tg4a Signals]
October 2005
Project: IEEE P Working Group for Wireless Personal Area Networks (WPANs)
Submission Title: [Spectral Properties of tg4a Signals]
Date Submitted: [30-Sep-2005]
Source: [Laurent Ouvry, Samuel Dubouloz, Mathieu Sambuq]
Company [CEA-Leti]
Address [17 rue des Martyrs Grenoble]
Voice:[ ], FAX: [ ],
E−Mail:
Re: [ a.]
Abstract: [Spectral Properties of tg4a Signals]
Purpose: [To promote discussion in a.]
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
L. Ouvry, S. Dubouloz, M. Sambuq — CEA-Leti

2. Objective Determine impact of :
October 2005
Objective
Determine impact of :
Peak PRF
Burst length
Scrambling (polarity and time)
Symbol duration (& mean PRF)
on signal average power (with FCC definitions)
Study of the signals peak power
Polarity and time Scrambling influence
Peak PRF (PRFmax)
L. Ouvry, S. Dubouloz, M. Sambuq — CEA-Leti

3. Scrambling polarity code word
October 2005
Introduction
Average Power Measurements
Measure in a 1MHz RBW Bandwidth
Averaging time lower or equal to 1ms (video integration time)
Use of matlab functions given in doc #354r1-UWB-Power-Measurements
Peak Power Measurements
Measurement in a 3MHz RBW
5MHz of Video Bandwidth (VBW)
Signal Models
Burst composed by N pulses repeated at peak “PRFmax”
Symbol time is noted Ts
Mean PRF is therefore N/Ts
Scrambling is used
Pulse polarity codes
Burst position scrambling
PPM Modulation
Fine time scrambling on the whole symbol time
FCC Compliant
Measurements
1 chip ~ (1/PRFmax) ns
N chips Ts
Scrambling polarity code word
L. Ouvry, S. Dubouloz, M. Sambuq — CEA-Leti

4. Average Measurements October 2005
L. Ouvry, S. Dubouloz, M. Sambuq — CEA-Leti

5. Worst Case : No polarity code, no time scrambling
October 2005
Worst Case : No polarity code, no time scrambling
Signal can be expressed as:
According to doc # 354r1:
Burst of N impulsions p(t)
Burst repeated every Ts
Pulse Spectrum composed by spectral lines
spaced by PRF and convolved by a sinc function of
N/PRF width TF
Pulse PSD, which max is dBm/MHz
RBW < 1/Ts
Spectral lines amplitudes, not depending on PRF
RBW ≥ 1/Ts
L. Ouvry, S. Dubouloz, M. Sambuq — CEA-Leti

6. Example (No polarity code, no time scrambling)
October 2005
Example (No polarity code, no time scrambling)
Measurement
Calculation
PRF = 247 MHz
Amplitude = *log10(RBW * N) – 60 =
Difference is due to sinc
Function sidelobes
L. Ouvry, S. Dubouloz, M. Sambuq — CEA-Leti

7. No Modulation – No Time Scrambling
October 2005
No Modulation – No Time Scrambling
Burst Size (N)
+ 3 dB
Y-scale is what is above –41.3dBm/MHz (in previous simple example, would be 8.6dB)
identical
L. Ouvry, S. Dubouloz, M. Sambuq — CEA-Leti

8. Modulation Effects PPM Modulation October 2005
It is worst with PPM modulation only
L. Ouvry, S. Dubouloz, M. Sambuq — CEA-Leti

9. Modulation Effects (2) PPM + Polarity October 2005
PPM + Polarity is better than PPM only
L. Ouvry, S. Dubouloz, M. Sambuq — CEA-Leti

10. Time Scrambling Effects
October 2005
Time Scrambling Effects
Fine time scrambling on the whole symbol time
 No Difference between simple modulation (2-PPM, 4-PPM) and fine time scrambling
L. Ouvry, S. Dubouloz, M. Sambuq — CEA-Leti

11. Polarity Codes Effect October 2005
Each pulse has a random polarity
Standard deviation is about 0.25dB
Polarity Codes are required for spectrum smoothing
L. Ouvry, S. Dubouloz, M. Sambuq — CEA-Leti

12. October 2005
Conclusions
Peak PRF has no influence on the average transmitted power (w.r.t. FCC)
Time scrambling has a minor effect on spectrum smoothing and is not better than PPM Modulation in this respect
Random Polarity codes are required as soon as bursts of pulses are used
Burst size and symbol time do not influence the spectrum anymore with random polarity codes
A power backoff (margin) of about 1.5dB is required  compliant with doc # 605r1
L. Ouvry, S. Dubouloz, M. Sambuq — CEA-Leti

13. Peak Measurements October 2005
L. Ouvry, S. Dubouloz, M. Sambuq — CEA-Leti

14. Peak Power Measurements
October 2005
Peak Power Measurements
Ts = 1µs
Generation of short signals (~10µs) and statistics (1000 realizations) on the peak power measurement, with random polarity scrambling sequences (length 8) and with/without time scrambling
The CDF of the difference between the measurement and the FCC Limit is plotted (if abscissa < 0, the measurement was below the limit and if > 0, the measurement was above the limit)
Measurement performed in a 3MHz RBW, so FCC limit is 20log10(3/50) = dBm
Below FCC Limit
Above FCC Limit
 No significant effect of the peak PRF  No significant effect of the time scrambling (performed randomly on whole Ts)  As for the average power, the biggest effect is given by the polarity scrambling (variation between -3 and +5 dB around the FCC peak Limit).
L. Ouvry, S. Dubouloz, M. Sambuq — CEA-Leti

15. Peak Power: Effect of polarity scrambling
October 2005
Peak Power: Effect of polarity scrambling
Generation of signal (~10µs) and peak power measurement, for all polarity scrambling sequences (with length 8  256 sequences, and symbol time = 1µs).
Some sequences enable a significant peak power reduction
Symmetry for
flip left  right
Complementary values -1  1
Example: code word 22 (0 -1 and left msb) is polarity scrambling sequence
Difference between the measurement and the FCC limit for all polarity sequences of length 8
(decimal)
Above FCC Limit A
FlipLR 22
233
Not(A)
151
104
Below FCC Limit
L. Ouvry, S. Dubouloz, M. Sambuq — CEA-Leti

16. Peak Power Best Polarity Scrambling Sequences
October 2005
Peak Power Best Polarity Scrambling Sequences
CDF for the previous curves
Best sequences depend on PRFmax
About 35% of polarity sequences provide a peak power lower than the FCC limit for a 494MHz PRFmax
About 50% for 247MHz
 New statistics with the best sequences only
Below FCC Limit
Above FCC Limit
L. Ouvry, S. Dubouloz, M. Sambuq — CEA-Leti

17. Peak Power Best Polarity Scrambling Sequences
October 2005
Peak Power Best Polarity Scrambling Sequences
Statistics on Peak Measurement (1000 realizations) but with only the polarity scrambling code words involving a peak measurement lower than the FCC limit
If there is no time scrambling, the peak measurement is always lower than the FCC limit (due to sequence choice)
If time scrambling, some occurrences are above the FCC limit
Difference between PRFmax 247MHz and 494 MHz is not significant
Below FCC Limit
Above FCC Limit
L. Ouvry, S. Dubouloz, M. Sambuq — CEA-Leti

18. Conclusions (Peak Power)
October 2005
Conclusions (Peak Power)
Very slight differences between 494MHz and 247MHz peak PRF
The choice of the polarity scrambling sequences has an important effect on the radiated peak power (but hardly compatible with simple scrambler…)
This choice does not affect the average power measurement
L. Ouvry, S. Dubouloz, M. Sambuq — CEA-Leti