Performance Investigation of Convolutional Vector Symbol Decoding with Larger than Two Choices and with Incomplete Second Choices

Abstract

Vector Symbol Decoding (VSD) is a decoding technique for both block and convolutional nonbinary encoders. Convolutional VSD uses a convolutional encoder with nonbinary symbols at the encoder and VSD at the decoder. VSD can employ diversity quite easily to improve the performance and simplify the decoding because diversity provides alternative choices for the decoder. Previous work always assumed two complete choices for simplicity. This paper explores the performance of a convolutional VSD for the case of up to four complete choices and the case of two incomplete choices. The results showed that 1) although the second choice clearly improved the performance, the third and fourth choices do not provide ample improvement to justify the higher complexity of the hardware. 2) VSD can use incomplete second choices to improve performance. In summary, second choices should be applied whenever possible, but no more than two choices should be used. These results are very useful for hardware implementation.