Characterizing Decoder Behaviour and Performance Limits Under Realistic Burst‐Error Conditions: A BD‐SHFMM‐Based Framework for NB‐PLC Systems

ABSTRACT

This paper presents a statistically grounded framework for characterizing decoder behaviour and performance limits of forward error correction (FEC) schemes in narrowband power line communication (NB‐PLC) systems under realistic burst‐error conditions. In such environments, impulsive and temporally correlated noise produces clustered errors, whereas conventional evaluations rely on simplified or memoryless channel models that fail to capture these dynamics and can overestimate performance. To address this limitation, block‐diagonal semi‐hidden Fritchman–Markov models (BD‐SHFMMs), trained on empirical NB‐PLC error traces, are used to generate statistically representative burst‐error sequences without full physical‐layer simulation. This enables controlled and reproducible analysis of decoder behaviour under realistic conditions. The framework evaluates Reed–Solomon (RS), convolutional and concatenated RS–convolutional coding schemes across multiple modulation formats and disturbance levels using bit error rate (BER), symbol error rate (SER) and error‐free run distribution (EFRD) metrics. The results reveal a fundamental divergence between average error performance and temporal error behaviour. While concatenated RS–convolutional coding achieves the lowest BER, it can reduce mean error‐free run length by more than 50% compared to convolutional coding under mild disturbance, indicating degraded temporal error separation. Furthermore, decoder performance exhibits threshold‐like behaviour, with abrupt degradation when burst lengths exceed the effective correction capability of the coding scheme, in some cases resulting in post‐FEC error rates worse than the uncoded baseline due to error propagation. These findings demonstrate that FEC performance in NB‐PLC systems is fundamentally governed by the burst‐error temporal structure, and that conventional scalar metrics alone are insufficient to capture decoder behaviour. The proposed BD‐SHFMM‐based framework provides a reproducible and practically meaningful tool for burst‐aware evaluation and the design of robust FEC strategies in NB‐PLC environments.