On Adaptive BICM with Finite Block-Length and Simplified Metrics Calculation

In this paper we present a novel performance analysis of bit-interleaved coded modulation (BICM) over AWGN channel for different modulation schemes considering a finite block length and suboptimal metrics calculation. Using the known BICM capacity analysis as well as the Gallager bound, we show that among many known modulations, only three M-ary quadrature amplitude modulation (QAM) are necessary to ensure the maximization of the channel capacity. We analyze the maximum throughput and the switching thresholds of the system, and we compare the theoretical analysis with simulation results of practical strong codes, i.e., turbo codes. We show that if the modulation and code are well adapted, the optimal throughput decreases by less than 1% when suboptimal metrics are used. We show also, that the turbo codes throughput is within 1.5 dB of the new bound, and the SNR switching thresholds, are shifted by 1.5 dB with respect to the theoretical analysis.