Author:

Nguyen Xuan Duc (Le Quy Don Technical University, Ha Noi, Vietnam), Pham Xuan Nghia (Le Quy Don Technical University, Ha Noi, Vietnam), Hoang Van Dung (Le Quy Don Technical University, Ha Noi, Vietnam)

Abstract:

The maximum likelihood sequence estimation (MLSE) equalizer achieves optimal performance over channels affected by inter-symbol interference (ISI); however, its complexity grows exponentially with the channel impulse response length. To address this, a block-wise MLSE approach combined with bit-interleaved coded modulation with iterative decoding (BICM-ID) is proposed, leveraging mutually independent signal blocks to enable parallel implementation, which significantly reduces computational complexity and processing latency for real-time applications. This paper analyzes and optimizes the precoding block length $n_2$ in the block-wise MLSE algorithm over ISI channel models with different impulse response lengths $L$. Simulation results show that the bit error rate (BER) performance of the system depends strongly on $n_2$, and that a saturation threshold exists. When $n_2 > 2L$, inter-block interference is effectively suppressed, allowing the BER performance to asymptotically approach that of the full-sequence MLSE. Conversely, when $n_2 \leq 2L$, residual inter-block interference leads to significant performance degradation or the appearance of an error floor. This quantitative design rule optimizes the precoding block length, effectively balancing performance and complexity for block-wise MLSE deployments in BICM-ID systems.