Citation/Abstract

Geometric perspectives of test-based decoding acceptance criteria are developed. Test-based decoding procedures employ simple algebraic tests conducted on candidate codewords, efficiently generated by low-complexity sub-optimal decoding strategies, and channel reliability measures. Test-based decoding procedures typically yield significant performance improvements over sub-optimal decoding procedures while maintaining low average complexity. Specific contributions are: (1) A geometric interpretation of a single codeword acceptance criterion for binary codes transmitted over a coherent additive white Gaussian noise channel provides the methodology to evaluate the performance of test-based decoding. (2) Test-based decoding procedures and acceptance criterion performance bounds are extended to other memoryless channels, the random phase and the Rayleigh fast fading channels. (3) Test-based decoding procedures and acceptance criterion performance bounds are extended to non-binary codes. (4) Single codeword test-based decoding procedures and acceptance criterion performance bounds are extended to multi-codeword test-based decoding, with specific emphasis placed on the two-codeword case.