Impact of Non-Square Modulation Formats on Multi-Layer Error- Correction Systems

Abstract

This work presents a set of algorithms for integrating non-square modulation formats with multi-layer forward error-correction (FEC) systems. This was done to improve the adaptive coding and modulation (ACM) capabilities of Ericsson’s nextgeneration digital signal processing (DSP) product. Specifically, support for cross- (X) quadrature amplitude modulation (QAM) formats with 9 to 19 bits per symbol was implemented. This was achieved while respecting limitations imposed by the multi-layer FEC, consisting of low-density parity check and Reed Solomon errorcorrecting codes. The main contributions were made to the symbol mapper and demapper as well as the most significant bit decoders. The algorithms were implemented in hardware using high-level synthesis, with the target platform being a field-programmable gate array (FPGA). Testing on the target FPGA was performed using a digital loop and an internal noise generator to produce bit error rate (BER) vs. signal-to-noise ratio (SNR) curves. The results displayed a 1 dB SNR gain when switching from a higher-order square QAM format to a lower-order X QAM format with higher code rate. Since these two options provide the same capacity, it implies that it is more beneficial to reduce modulation order to improve BER performance compared to increasing FEC overhead when channel conditions degrade. Hence, the implementation of X QAM formats has the potential to improve the ACM flexibilty in Ericsson’s DSP product, with only a small increase in hardware usage.