Coding and Information Theory Group
The Coding and Information Theory Research Group (CIT) brings together researchers interested in the theory and applications of efficient storage and transmission of information.
Overview
|
The Coding and Information Theory Research Group is a forum for research into the fundamental theories of information transmission (mainly at the physical layer). Our strengths and interests are in the areas of:
|
Group Leader
Project examples
Decoder Scheduling in Turbo Decoders
An interesting practical consideration for decoding of serial or parallel concatenated codes with more than two components is the determination of the lowest complexity component decoder schedule which results in convergence. Work performed in collaboration with researchers from Chalmers University of Technology, Sweden has resulted in an algorithm that finds such an optimal decoder schedule. A technique was also found for combining and projecting a series of three-dimensional extrinsic information transfer (EXIT) functions onto a single two-dimensional EXIT chart. This is a useful technique for visualizing the convergence threshold for multiple concatenated codes and provides a design tool for concatenated codes with more than two components.List detection for APP Approximation
The APP decoder is a fundamental building block of iteratively decoded systems. In many cases of interest, exact computation of posterior probabilities is computationally infeasible. Examples include multiple-access channels and multi-antenna channels. ITR researchers have pioneered low-complexity methods for approximating posterior probabilities via the use of list detection. These algorithms find a small list of sequences with sufficiently high a-posteriori probability and approximate the a-posteriori probability computation by marginalisation over these sequences, rather than all possible sequences. This leads to practical, high performance iterative decoders.Transmit Beamforming
Transmit beamforming is a simple technique to improve the received signal-to-noise ratio on fading channels. Using the theory of random matrices, a performance analysis was found for transmit beamforming over correlated Rayleigh fading channels. In particular, exact finite antenna expressions were found for the average bit error rate (in the case of ergodic channels) for both non-coherent and coherent detection. Expressions for the outage probability (in the case of quasi-static channels) were also determined. Part of this work was performed in collaboration with researchers from the CSIRO ICT Centre.MIMO Capacity
Shannon theoretic results have been obtained for multiple-input multiple-output channels with arbitrarily distributed, correlated fading. In conjunction with researchers from National ICT Australia, an optimality condition was found for the input covariance for ergodic Gaussian vector channels with arbitrary channel distribution. Using this optimality condition, an iterative algorithm was found for numerical computation of optimal input covariance matrices. This work has application to correlated Rayleigh and Ricean channels, where the effect of correlation and line-of-sight paths can now be evaluated.Space-Time Code Design
Most space-time codes are designed to minimize a certain upper bound on the bit error rate. However in many cases of interest, it is the frame error rate that is of more importance – particularly for packet transmission over slow fading channels. ITR researchers have designed space-time codes specifically to minimize frame error rate on such channels. The resulting codes outperform all previously known codes.Error Control Coding Theory
The performance of iterative decoding can be limited by the presence of cycles in the graphical structure of the code. In particular, cycles of length four have a detrimental effect on performance. In collaboration with researchers from the University of Southern California, a relationship between the minimum distance, dimension and block length of codes was discovered, which is required for the absence of such four-cycles. Using this relationship, it was shown that many classical codes, such as Hamming, Reed-Solomon, BCH, Reed-Muller, and Golay codes have unavoidable four-cycles in their graphical representation.Contact
| Professor Alex Grant Research Professor of Information Theory |
|
| Alex.Grant@unisa.edu.au | |
| Phone | +61 8 8302 5219 (Office)
+61 0417887914 (Mobile) |
| Fax | +61 8 8302 3873 |
| Post |
Room 2-14
Institute for Telecommunications Research Mawson Lakes Boulevard Mawson Lakes SA 5095 AUSTRALIA |