• Summary
• Related Documents
  • Active Bibliography
  • Co-citation
• Version History

Abstract:

Limits for the performance of communication systems can, in principle, be described by what is called OPTA (optimal performance theoretically attainable) curves. Approaching these limits necessitates large delays and complexities. It can be derived from Shannon's works [22, 23] that the limit can be reached by separately optimizing the source- and channel coders. For existing practical systems the performance is not close to the OPTA curve. And there is strong evidence that combined source- and channel coding at a given complexity level and allowing for a maximum delay will improve the systems, both in terms of rate-distortion performance and robustness towards variable channel conditions. In this article we propose a novel technique where analog source representation vectors are approximated and mapped to analog channel vectors with reduced dimensionality. Sample rate reduction implies compression at a system level. To evaluate the performance of such a system we have defined compression as the ratio between the sample rate of a reference system without compression and the sample rate of the proposed system when both offer equal received signal quality. The robustness exhibited by the proposed method, which is especially valuable for wireless systems where the channel conditions vary over a large range, is obtained by neighboring relationships between channel vectors and source vectors. The proposed method performs better than any other known method and is close to the OPTA curves for the synthetic sources tested. For practical communication applications it gives far stronger robustness towards channel variations than earlier proposed methods.

Citations