University of Delaware - College of Engineering


Communication & Signal Processing

Channel coding for non-standard channels with memory

Javier Garcia-Frias

Traditional error correcting codes fail when the errors in the channel occur in bursts (or with memory). In order to deal with this problem, the standard approach is to use a channel interleaver so that the data is shuffled before transmission through the channel. By using a channel deinterleaver before the decoding process at the receiver, the errors appear to the decoder in a random manner and therefore traditional schemes are able to achieve some error protection. However, it is theoretically well know that destroying the memory in the errors degrades the achievable performance. Thus, our objective is to design practical codes capable of achieving the gains promised by the theory. In order to achieve this goal, we have introduced a technique to exploit the channel memory in turbo decoding when this memory is described by hidden Markov models, achieving, for the first time, a performance that is better than traditional schemes using channel interleavers and ignoring the channel memory. This method can be successfully applied in real channels by estimating the channel model jointly with the decoding process. The application of these techniques in real channels such as DECT leads to a significant performance improvement with respect to standard systems.

Recent publications

H. Lou and J. Garcia-Frias, "Low-Density Generator Matrix Codes for Indoor and Markov Channels", submitted to IEEE Trans. on Wireless Communications.

R. D. Souza and J. Garcia-Frias, "Semi-Blind Iterative Markov Receiver for Space-Time Coded Modulation over Time-Correlated Fading Channels", submitted to IEEE Trans. on Wireless Communications

R. D. Souza, H. Lou, and J. Garcia-Frias, "On the Effect of Fading Correlation in the Performance of an Iterative Receiver for Quasi-Static ISI MIMO Channels", CISS 05, March 2005, Baltimore, Maryland.

Bookmark and Share