Prof. Brian L. Evans

Dept. of Electrical and Computer Engineering

Wireless Networking and
Communications Group

The University of Texas at Austin, Austin, Texas

Lead graduate students: Mr. Aditya Chopra, Mr. Kapil Gulati and Mr. Marcel Nassar

Current collaboration with Dr. Eddie Xintian Lin, Dr. Alberto Alcocer Ochoa, Dr. Kathyayani Srikanteswara and Mr. Keith R. Tinsley at Intel

Tuesday, April 12, 2010

Slides in PowerPoint 2007 format and PowerPoint 2003 format

Wireless transceivers are affected by radio frequency interference (RFI) generated from nearby electronic devices (e.g. microwave ovens), coexisting wireless communication sources, and computational platform clocks/busses. RFI is well modeled using non-Gaussian impulsive statistical distributions and can severely degrade the communication performance of wireless transceivers designed under the assumption of additive Gaussian noise. The problem intensifies with higher reuse of radio spectrum and shrinking form factor of the computational platform.

In the first part of the talk, we present our results on statistical modeling and mitigation of RFI in wireless receivers. In particular, we first establish the applicability of the Symmetric Alpha Stable, Middleton Class A, and Gaussian mixture distributions to model RFI in various interference scenarios. Scenarios include Wi-Fi, Wimax, cellular, ad hoc, and wireless sensor networks. Using these statistical models of RFI, we discuss several filtering and detection methods to mitigate RFI for single- and two-antenna receivers. We demonstrate 1-2 orders of magnitude reduction in bit error rate for the same transmission rate, and evaluate design tradeoffs of our proposed methods. RFI modeling also has the potential to improve communication performance by 1-2 orders of magnitude at the medium access control layer.

In the second part of the talk, we demonstrate our freely distributable RFI modeling and mitigation toolbox. Our toolbox can be used to design RFI immune transceivers using easy-to-use GUI tools built on top of RFI modeling and mitigation algorithms. The toolbox can be used by a system design engineer for platform analysis/design and a communications engineer for wireless network performance analysis. We how to use the toolbox to

- (a) generate RFI noise/interference,
- (b) fit measured RFI data in an embedded laptop receiver to statistical models
- (c) quantify impact of RFI on standard receivers, and
- (d) demonstrate communication performance improvement using RFI mitigation techniques.

http://users.ece.utexas.edu/~bevans/projects/rfi/software/index.html

This research has been supported by Intel since January 2007.

Agenda:

(30 min) Presentation

(20 min) RFI Modeling and Mitigation Toolbox Demo

**Answer #1**.
The myriad filter for pre-processing has a single parameter
whose optimal value is computed as follows:

Here, alpha is the exponent in the Symmmetric Alpha Stable distribution where 0 < alpha < 2, and y is the dispersion parameter (analogous to variance).

Here are the values of derivation of k(alpha) with respect to alpha for selected values of alpha:

alpha k'(alpha) ----- --------- 0.5 0.7698 sqrt(y) + 0.57735 sqrt(y) ln(y) 1.0 y + y ln(y) 1.5 2.31 y^(3/2) + 1.73205 y^(3/2) ln(y)For y = 1, k'(alpha) varies from 0.7698 to 2.31 for 0.5 < alpha < 1.5. In fitting RFI data, we have found that alpha > 0.5. Pertubations in k are particularly severe as alpha -> 0 and as alpha -> 2.

*Question #2*. In your presentation, you showed that
the Myriad prefilter can mitigate RFI with 10 samples/symbol.
Can the Myriad prefilter still mitigate RFI with fewer samples/symbol?

**Answer #2**.
Yes, Myriad pre-filtering can still improve communication
performance. The size of the sliding window over which the Myriad
filtering operation is done has to be modified appropriately.
The number of samples in the sliding window should generally be
smaller than the number of samples per symbol.
The performance improvement is larger for a higher number of
samples per symbol.
In alpha stable noise with alpha=0.8, the symbol error rate
decreases by two orders of magnitude for the Myriad filter
vs. the matched filter for 4 samples/symbol, 3 samples/window,
and 3 dB SNR in simulations.

Mail comments about this page to bevans@ece.utexas.edu.