Embedded Signal Processing Laboratory

Department of Electrical and Computer Engineering

The University of Texas at Austin, Austin, Texas USA

01/22/13

Interference Modeling and Mitigation Research at UT Austin

Translations:

German,
courtesy of Katia Osipova.

Polish, courtesy of Alexey Gnatuk.

Romanian, courtesy of azoft.

Russian, courtesy of
Oleg Meister.

Slovak, courtesy of
Sciologness Team

The toolbox enables a user to

- generate impulsive noise/interference
- fit measured data to impulsive noise models
- apply nonlinear filters to denoise a signal in impulsive noise
- improve detection performance of a signal in impulsive noise

In communication systems, we model asynchronous interference as additive impulsive noise. We've derived the impulsive noise distributions by using statistical-physical models of propagation:

- Gaussian mixture model for the case when the receiver is located in a spatial guard zone, e.g. in cellular, Wimax, and Wi-Fi networks.
- Symmetric alpha stable distribution when there is no guard zone, e.g. wireless sensor networks.

Impulsive noise is also a problem in wireline communications. In powerline communication systems, for example, impulsive noise is the dominant noise component. The impulsive noise arises from many sources such as switching circuitry and external transmissions. We have shown that the impulsive noise distribution follows a Gaussian mixture model. In certain cases, the Gaussian mixture model simplifies to a Middleton Class A model.

When communication receivers are designed assuming that the only additive noise source is Gaussian spectrally-flat noise, the presence of additive impulsive noise can cause severe degradation in timing recovery, frame synchronization, equalization, detection and error correction subsystems. Conversely, by redesigning the receiver with additive impulsive and thermal noise in mind, the receiver can achieve up to 20 dB of SNR gain in the presence of impulsive noise. We have achieved this improvement by adding a pre-filter, or changing the detector, or adapting a Turbo decoder. An SNR gain of 20 dB would translate into a significant reduction in bit error rate (by 1-2 orders of magnitude) [10] or improvement in bit rate (of up to 3-6 bits/s/Hz) in interference-limited channels. We have used RF statistics to derive the tradeoffs in throughput, delay and reliability and demonstrate the throughput can be doubled and reliability improved over an existing medium access layer protocol [11].

- Symmetric Alpha Stable model
- Middleton Class A model
- Gaussian Mixture model

The Middleton Class A model is a special case of Gaussian mixture.

We have used the toolbox to fit radio frequency interference data measured at baseband [6][7][8][9].

The toolbox also generates the following time series:

- Multivariate isotropic Symmetric Alpha Stable model
- Multivariate isotropic Middleton Class A model
- Multivariate Gaussian mixture model

The toolbox implements several impulsive noise reduction filters:

- Middleton prefiltering [1]
- Myriad filtering [5]

The toolbox supports single-transmitter single-receiver (1x1) systems, as well as two-transmitter, two-receiver (2x2) systems. For 1x1 communication systems, the toolbox uses pulse amplitude modulation (PAM). Receiver choices are correlation detection, Wiener filtering followed by correlation detection, optimal Bayes detection [1], and the small-signal approximation of optimal Bayes detection [8]. For 2x2 communication systems, the toolbox uses quadrature amplitude modulation (QAM). Transmission uses either spatial multiplexing or Alamouti coding. Reception choices include Gaussian maximum likelihood (ML), zero forcing, Middleton Class A ML, and suboptimal Middleton Class A ML receivers. [6]

- Interference Modeling and Mitigation Toolbox 1.6,
Apr. 1, 2011.
The licensing terms for the toolkit are now far more flexible with version 1.6. Version 1.6 adds

- functions to generate multidimensional (isotropic) data from a symmetric alpha stable distribution (RFI_MakeDataAlphaStable) and Middleton Class A distribution (RFI_MakeDataClassA, RFI_MakePDFClassA, and RFI_MakeEnvelopeDataClassA),
- a demonstration for analyzing communication performance of multicarrier receivers in the presence of RFI (RFI_OFDMDemo),
- help information when placing the cursor over a text field in a demonstration, and
- help buttons in the demonstrations.

Version 1.6 requires the following:

- Matlab 7.0 or higher
- Communication Toolbox (for pammod, pamdemod, qammod, qamdemod, and rcosine functions)
- Signal Processing Toolbox (for xcorr function)
- Statistics Toolbox (for ksdensity, moment, mvnpdf, mvnrnd, random, and randsample functions)

Here, please replace 'C:\Program Files\MATLAB\R2007a\toolbox\' with the destination directory to where you had copied the rfitoolbox folder.

** Note**: Starting with version 1.3, a main GUI demo has
been available to run all other demos included in the release.
The main demo can be started by typing 'RFI_StartDemos' on the MATLAB
command prompt after completing the aforementioned installation procedure.

- Report (May 2007)
- Presentation (Oct. 2009)

- RFI Modeling and Mitigation Toolbox 1.5,
Aug. 16, 2010.

Version 1.5 adds generation of random variables/processes and parameter estimation for the Gaussian mixture model (GMM). Here are selected improvements for version 1.5:- Added RFI_MakePDFGMM.m to generate a Gaussian mixture model probability density function
- Added RFI_MakeDataGMM.m to generate Gaussian mixture model data
- Added RFI_EstParamsGMMwithEM.m to fit a given vector of time series samples to a Gaussian mixture model.
- Added generation of impulsive noise data from multivariate Gaussian mixture models
- Added Gaussian mixture model to the Statistical Modeling demo.
- Added partial support for Gaussian mixture models to the single-input single-output (SISO) demo.
- Merged the file transfer demo and the SISO demo. As a result, the SISO demo now contains the ability to specify a file as a source of the simulation data.
- Added tips and help for fields that show upon moving the mouse on top of the corresponding fields.

- Matlab 7.0 or higher
- Communication Toolbox (for pammod, pamdemod, qammod, qamdemod, and rcosine functions)
- Signal Processing Toolbox (for xcorr function)
- Statistics Toolbox (for ksdensity, moment, mvnpdf, mvnrnd, random, and randsample functions)

- RFI Modeling and Mitigation Toolbox 1.4.1 beta,
Apr. 11, 2010.

Version 1.4.1 beta simply adds a new overview demonstration called RFI Modeling and Mitigation in Single-Input Single-Output (SISO) Systems. Otherwise, version 1.4.1 beta is the same as version 1.4 below. - RFI Modeling and Mitigation Toolbox 1.4,
Feb. 6, 2010.

Here is a summary of improvements for version 1.4:- Updated the demo RFI_DemoSISO to support M-PAM modulation (previous versions supported only 2-PAM).
- Updated the function RFI_MakeDataAlphaStable to support generation of skewed alpha stable random variables.
- Fixed a minor bug in the function RFI_CalcKLDiver for robustness in calculating KL divergence.

- Matlab 7.0 or higher
- Communication Toolbox (for pammod, pamdemod, qammod, qamdemod, and rcosine functions)
- Signal Processing Toolbox (for xcorr function)
- Statistics Toolbox (for ksdensity, moment, random and randsample functions)

- RFI Modeling and Mitigation Toolbox 1.3,
Aug. 26, 2009.

Version 1.3 requires the following:- Matlab 7.0 or higher
- Communication Toolbox (for pammod, pamdemod, qammod, qamdemod, and rcosine functions)
- Signal Processing Toolbox (for xcorr function)
- Statistics Toolbox (for moment, random, randsample and ksdensity functions)

- Added a demo (RFI_DemoStatisticalModeling) which provides a simulation environment for statistical modeling of RFI.
- Added a demo (RFI_StartDemos) which is the main demo from which all other demos can be launched.
- Added function to evaluate the probability density function of a bivariate Middleton Class A random variable (RFI_MakePDFBiVarClassA).
- Added a function (RFI_kde2d) which performs 2-D kernel density estimation. Copyright information regarding the redistribution of this function has been added as comments in the function.
- Added a function to evaluate the Kullback-Leibler (KL) divergence between two 1-D or 2-D probability density functions (RFI_CalcKLDiver).
- Added a short help and description file for all existing demos in this toolbox.
- Fixed a typographical bug in the demo RFI_DemoTwoByTwoMIMO.
- Removed the sample .avi file (einstein.avi) that was present in releases 1.2 and 1.2.1.

- RFI Mitigation Toolbox 1.2.1 beta,
Apr. 3, 2009.

Fixes a bug in the function to generate Middleton Class A noise, RFI_MakePDFClassA. - RFI Mitigation Toolbox 1.2, Feb. 7, 2009.

Version 1.2 requires the following:- Matlab 7.0 or higher
- Communication Toolbox
(for
`pammod`

,`pamdemod`

,`qammod`

,`qamdemod`

, and`rcosine`

functions) - Signal Processing Toolbox
(for
`xcorr`

function) - Statistics Toolbox
(for
`moment`

,`random`

and`randsample`

functions)

- Added functions for generation of bivariate Middleton Class A noise.
- Modified the Middleton Class A noise generators for improved computational performance.
- Added functions to implement 2x2 MIMO receivers in the presence of Gaussian and bivariate Middleton Class A noise.
- Added demo for a 2x2 MIMO system in the presence of RFI (RFI_DemoTwoByTwoMIMO)
- Added small signal approximation and quantized pdf implementation of the Bayesian detection in the presence of Middleton Class A noise
- Added demo for single-carrier transmission and reception of a file in the presence of either additive symmetric alpha stable noise or additive Middleton Class A noise

- RFI Mitigation Toolbox 1.1 beta, Nov. 21, 2007.

Version 1.1 beta requires the following:- Matlab 7.0 or higher
- Communication Toolbox
(for
`pammod`

and`rcosine`

functions) - Signal Processing Toolbox
(for
`xcorr`

function) - Statistics Toolbox
(for
`moment`

function)

- Added myriad filtering [5] support
- Added demo for communication performance in alpha stable noise
- Fixed dispersion parameter calculation to agree with [2]
- Improved speed of the Middleton Class A noise generator
- Added recursive implementation of the Middleton Class A PDF estimator to improve speed
and increase the usable range of the overlap index parameter
*A*

- RFI Mitigation Toolbox 1.0, Sept. 22, 2007.

Version 1.0 requires the following:- Matlab 7.0 or higher
- Communication Toolbox
(for
`pammod`

and`rcosine`

functions) - Signal Processing Toolbox
(for
`xcorr`

function) - Statistics Toolbox
(for
`moment`

function)

- A. Spaulding and D. Middleton, "Optimum reception in an impulsive
interference environment-part I: Coherent detection",
*IEEE Transactions on Communications*, vol. 25, no. 9, pp. 910-923, 1977. - G. A. Tsihrintzis and C. L. Nikias, "Fast estimation of the
parameters of alpha-stable impulsive interference",
*IEEE Transactions on Signal Processing*, vol. 44, no 6, pp. 1492-1503, June 1996. - D. Middleton, "Procedures for determining the properties of the
first-order canonical models of Class A and Class B electromagnetic
interference",
*IEEE Transactions on Electromagnetic Compatibility*, vol. 21, pp. 190-208, Aug. 1979. - S. M. Zabin and H. V. Poor,
"Efficient estimation of Class A noise parameters
via the EM [Expectation-Maximization] algorithms",
*IEEE Transactions on Information Theory*, vol. 37, no. 1, pp. 60-72, Jan. 1991. - J. R. Gonzalez and G. R. Arce.
"Optimality of the myriad in practical impulsive-noise environments,"
*IEEE Transactions on Signal Processing*, vol. 49, no. 2, pp. 438-441, Feb. 2001. - K. Gulati, A. Chopra, R. W. Heath, Jr., B. L. Evans, K. R. Tinsley,
and X. E. Lin,
"MIMO Receiver Design in the Presence of Radio Frequency Interference",
*Proc. IEEE Int. Global Communications Conf.*, Nov. 30-Dec. 4th, 2008, New Orleans, LA USA. - M. Nassar, K. Gulati, A. K. Sujeeth, N. Aghasadeghi, B. L. Evans and
K. R. Tinsley,
"Mitigating Near-Field Interference in Laptop Embedded
Wireless Transceivers",
*Proc. IEEE Int. Conf. on Acoustics, Speech, and Signal Proc.*, Mar. 30-Apr. 4, 2008, Las Vegas, NV USA. - M. Nassar, K. Gulati, M. R. DeYoung, B. L. Evans and K. R. Tinsley,
"Mitigating Near-Field Interference in Laptop Embedded Wireless Transceivers",
*Journal of Signal Processing Systems*, Mar. 2009, invited paper. - K. Gulati, A. Chopra, B. L. Evans, and K. R. Tinsley,
"Statistical Modeling of Co-Channel Interference",
*Proc. IEEE Int. Global Communications Conf.*, Nov. 30-Dec. 4, 2009, Honolulu, Hawaii. - K. Gulati, B. L. Evans, J. G. Andrews and K. R. Tinsley,
"Statistics of Co-Channel Interference in a Field of
Poisson and Poisson-Poisson Clustered Interferers",
*IEEE Transactions on Signal Processing*, vol. 58, no. 12, Dec. 2010. - K. Gulati, R. K. Ganti, J. G. Andrews, B. L. Evans and
S. Srikanteswara, "Local Delay and Throughput-Delay-Reliability
Tradeoff in Wireless Ad Hoc Networks with Temporally Dependent
Interference",
*IEEE Transactions on Signal Processing*, to be submitted.

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