Abstract

This talk examines the design of a bit-interleaved coded-modulation with iterative decoding (BICM-ID) system for transmission over a quasi-static fading (QSF) channel. We illustrate the design process for matching the extrinsic information transfer (EXIT) chart curves for 16-QAM mapping schemes and memory-two recursive systematic convolutional (RSC) codes.

This design process is extremely complex and we define two classification metrics that enable us to reduce the number of mapper and encoder combinations. We use the probability of a decoding erasure to separate all 22 memory-two RSC codes into five groups. We conjecture that the codes in each group will have the same FER performance when coupled with the same mapping scheme. Similarly, we conjecture that all mapping schemes with the same distance spectrum will have identical FER performance curves when coupled with the same encoding scheme. We provide evidence to support the validity of these conjectures, which allow us to make the set of 16-QAM mapping schemes practically searchable when attempting to find the best match for a particular encoder.

Finally, we demonstrate the practical methodology for finding the best FER performance that can be achieved by a BICM -ID scheme using a memory-two RSC code matched to a 16-QAM mapping scheme for transmission over a QSF channel with various levels of antenna diversity. For example, by using our design framework, we observe that we are able to achieve a gain of 0.2 dB at a FER of 2×10^-2 and a gain of 0.4 dB at a FER of 4×10^-4 in the low and the high antenna diversity scenarios respectively over the RSC memory-two code (1, 5/7) coupled with the best mapping schemes found in the literature.