Source-matched channel coding and networking techniques for mobile communications

Abstract

The bit-stream generated by most source coding algorithms, for voice, audio and images, usually consists of bits of different significance and error sensitivity. By exploiting this property, in the design of a wireless communication system, it is possible to obtain significant advantages. First, the network capacity can be increased, with a smoother quality degradation under heavy traffic or severe interference conditions. Secondly, the application of source sensitivity matched FEC (forward error correction) protection schemes, leads to a higher robustness of source bits to channel impairments. We evaluate the usefulness of prioritized, variable-rate speech encoding schemes, for handling congestion over packet switching systems. The system performance is shown to be sensitive to the fraction of information placed in each priority and to the number of priorities. The applicability of source-sensitivity matched error protection schemes, for variable-rate speech transmission on a radio channel, is also investigated.