Coexistence of Multiband OFDM and IEEE 802'11a: Interference Measurements

1
Project IEEE P802.15 Working Group for Wireless
Personal Area Networks (WPANs) Submission Title
Multi-Band OFDM Interference on In-Band QPSK
Receivers Date Submitted 13 July,
2004 Source Celestino A. Corral, Shahriar
Emami, Gregg Rasor Company Motorola Address
8000 W. Sunrise Blvd., Plantation, Florida, USA
33322 Voice954-723-3864, FAX
954-723-3883 Re Abstract This document
provides simulation and theoretical results that
demonstrate MB-OFDM is an extremely harmful type
of interference to wideband in-band QPSK systems
such as TVRO receivers. A MB-OFDM interference
model is derived based on simulation and
analytical results. Purpose For discussion by
IEEE 802.15 TG3a. 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.
2
Multi-band OFDM Interference on In-Band QPSK
Receivers

• Celestino A. Corral, Shahriar Emami and Gregg
  Rasor
• Motorola
• 8000 W. Sunrise Blvd.
• Plantation, Florida
• July 13, 2004

3
Motivation

• Goal To characterize the impact of Multi-band
  OFDM UWB interference on in-band broadband
  wireless system like C-band satellite receivers.
• Note Multi-band OFDM (MB-OFDM) and Multi-band
  UWB (MB-UWB) requires power scaling of the
  waveform to compare competing technologies based
  on interpretation of FCC rules.
• Model of MB-OFDM interference derived. This
  model is bounded by periodically gated AWGN and
  impulsive MB-OFDM interference.
• Reconcile observed test results of MB-OFDM
  interference on satellite receivers as presented
  in ABQ meeting.

4
Multi-band UWB Power

• FCC states power spectral density for UWB devices
  must be -41.2 dBm/MHz in band between 3.1 and
  10.6 GHz
• Since multi-band signals hop over a selected band
  of frequencies, the power spectrum is scaled by
  the hop and averaged over the band.
• The resulting power spectral density is made
  equal to a system over any arbitrary band.

Multi-band spectrum
PSD level
f1
f2
fx
Integrate the spectrum over band and average by
band
To implement equal PSD over hop bandwidth, we need
requiring a power scaling.
5
Multi-band UWB Power
Equate power
Both systems transmit with equal power at a given
range.
Actual MB-OFDM PSD over its transmission
bandwidth.
Assuming DS-UWB bandwith is 2 GHz and MB-OFDM
bandwidth is 528 MHz.
6
In-band Receiver Filters
BW lt 40 MHz
High-Q band-pass filter can be approximated by
1
Band-pass Filter Frequency Response
Q gt 50 typical
complex frequency of band-pass filter
complex frequency of low-pass prototype filter
fc gt 3 GHz
Step response of band-pass filter has low-pass
impulse response envelope
Temporal characteristics of high-Q band-pass
filter determined by low-pass prototype. This
includes rise time, which obeys the following
relation 2
Not a function of filter approximation or order
Rise time of band-pass filter determined by 3dB
bandwidth of low-pass prototype.
1 A. Papoulis, The Fourier Integral and its
Application, Chap. 7, New York McGraw-Hill,
1962. 2 G. E. Valley, Jr., and H. Wallman,
Vacuum Tube Amplifiers, New York McGraw-Hill,
1948.
7
In-band Receiver Filters
Band-pass filter rise time for 40 MHz bandwidth.
Filter responds quite fast and observes virtually
full power of filtered MB-OFDM signal.
Filter with slower response.
Portion due to filter bandwidth
Portion due to temporal response
Received power
in dB
8
Coherent Detection QPSK Simulation
Noise Source
Block Diagram of Simulator
Matched Filter
Window
Detector

• QPSK system at 27.05 Msym/sec, similar to Dubai
  EDTV at 4020 MHz.
• 0 lt Eb/No lt 30 dB.
• 1000 symbols, 500 packets per Eb/No set.
• Sample rate 120 samples/QPSK symbol.
• Multi-band OFDM and all gated noise is 896
  samples long.

• Assume perfect synchronization
• Assume perfect phase estimation
• Input filter bandwidth wide enough so rise time
  not a factor
• Interference bandwidth is very large relative to
  filter bandwidth and approaches thermal noise as
  in 3.

3 J. Brandao, Interference effect on the
performance of PSK and QAM systems, IEE
Proceedings I, vol. 138, pp. 331337, Aug. 1991.
9
Simulation Results Gated Noise
3 dB
10
Simplified Theoretical Reason
Probability of symbol error for QPSK 4
Q-function for communication
Gated noise duty cycle Np is time interference
is present, Ns is time interference is silent.
Probability of error is due only to when the
noise is present Pep for the case it is silent
Pes 0
Actual error must be scaled by duty cycle as this
is time interference is present
equivalent quasi-fading of bit energy relative
to fixed noise power No
Q is very sensitive to r under high
signal-to-noise (SNR), meaning small changes in
duty cycle will impact probability of error when
minor changes in bit energy is most significant.
4 J. G. Proakis, Digital Communications, 4th
Ed., Boston, MA McGraw-Hill, 2001.
11
Theoretical vs. Simulated Results Gated Noise
Simulated
12
Simulation Results MB-OFDM
3 hops
AWGN
9 dB
13
Simulation Results Impulsive MB-OFDM
Worst-case peak-to-average power assumed for each
MB-OFDM symbol
Theory
?? dB
14
Simulation Results 3 hops
Impulsive MB-OFDM upper bound
MB-OFDM
Gated AWGN lower bound
15
MB-OFDM Interference Model
Amplitude distribution of AWGN
Amplitude distribution of MB-OFDM
Amplitude distribution of Impulsive MB-OFDM
MB-OFDM Model is Gaussian and Impulsive

• Multi-band OFDM transmissions can be long or
  bursty
• Long transmissions have amplitude distribution
  that approaches AWGN
• Bursty transmissions can be potentially
  impulsive
• We need to combine the Gaussian and impulsive
  characteristics

16
MB-OFDM Interference Model
Interference has gaps in time i.e., non-zero
probability of time during which there is no
interference in the receiver.
Class A Model 5
Interference time
Receiver bandwidth
Carrier-to-noise ratio
Peak factor (PAP)
Model Incorporates Gaussian and Impulsive Factors
for M-ary QAM
impulse index
Model
std. dev.
mean power ratio
Average symbol error rate
5 D. Middleton, Non-Gaussian noise models in
signal processing for telecommunications New
methods and results for class A and class B noise
models, IEEE Trans. Inform. Theory, vol. 45, pp.
11291149, May 1999.
17
MB-OFDM Model
Model
Simulated
18
Filtered MB-OFDM
40 MHz
9 subcarriers
The filtered waveform is generated and then
scaled to obtain same power as AWGN over the
packet. The waveform is then scaled by a
factor of 9/128 (in number of subcarriers) to
reduce the level to a filtered amount. This is
almost the same amount as 40/528 (in MHz), which
corresponds to the desired power reduction
relative to the full bandwidth of the
128-subcarrier symbol
9/128 factor (40/528 factor)
Assumed 0 dB
Received power
19
Simulation Results Filtered MB-OFDM
20
Conclusions

• Multi-band OFDM and Multi-band UWB equate power
  spectral density by scaling power in the hop and
  averaging over the entire hop bandwidth. This
  equates the transmitted power of a Multi-band
  system with DS-UWB over a fixed bandwidth.
• Probability of symbol error shows gated noise is
  akin to quasi-fading of bit energy relative to
  fixed AWGN level.
• The gated and scaled interference is more harmful
  than AWGN depending on the hop depth. Gated
  noise interference produces performance 3 dB from
  theory MB-OFDM produces performance 8 dB from
  theory.
• Multi-band OFDM can be impulsive. Under
  worst-case peak-to-average power Multi-band OFDM
  is a significant interferer to in-band coherent
  detection QPSK receivers.
• MB-OFDM model was derived based on combination of
  Gaussian and impulsive characteristics of MB-OFDM.

21
Narrowband Filter Response
Only upper portion of response captured
Wideband Filter Response
Narrowband Filter Response

• Fast rise time
• Delay applies across entire response
• Full level of interference reached within
  response time of the filter, and present for most
  of the interference time.
• Total power captured

• Slow rise time
• Delay applies across entire response
• Full level of interference not reached within
  response time of the filter.
• Total power can be captured if rise time and
  interference time are about equal.

Narrowband filters favor narrow pulsed
interference full level of interference is not
captured.
22
Backup Gated Noise Results for Other Hops
Increasing hop depth results in more degradation
at high SNR.
13
3
7
23
Backup MB-OFDM Results for Other Hops
Increasing hop depth results in more degradation
at high SNR.
13
3
7
24
Backup Impulsive MB-OFDM Results for Other Hops
Increasing hop depth results in more degradation
at high SNR.
3
13
7
25
Backup Gated Noise Theoretical Results for Other
Hops
13
3
7
26
Backup MB-OFDM Class A Model Results for Other
Hops
27
Properties of Q and Quasi-Fading
Working with Q(x) directly is difficult. We use
approximation
where 4
as then so
decays more rapidly
4 P. L. Borjesson and C-E. W. Sundberg, Simple
approximations of the error function Q(x) for
communication applications, IEEE Trans. Commun.,
vol. COM-27, pp. 639643, March 1979.
28
Peak-to-Average Power Tracking
Peak-to-average of AWGN and MB-OFDM track over
different hop depths.