Sparse MLD Decoder for 1-Bit ADC MIMO Constant Envelope Modulation

Due to the power inefficiency of the MIMO-OFDM components (i.e. linear power amplifier and high resolution ADC), the MIMO constant envelope modulation (MIMO-CEM) with 1-bit ADC was introduced as a low power communication system. However, the received MIMO-CEM signal suffers from an enormous distortion as a result of using the 1-bit ADC at MIMO-CEM receiver side (RX). Therefore, a very high complex IF based maximum likelihood decoder (IF-MLD) is used with the MIMO-CEM RX to neutralize the 1-bit ADC quantization effect. In the IF-MLD the whole MIMO-CEM is replicated in the IF band in each MLD state, which is too complex to be practical. Therefore, an efficient low complex kernel sparse MIMO-CEM MLD is proposed. Where, a linear baseband MLD is proposed based on simple yet efficient approximation for the 1-bit ADC. Then, a kernel sparse selector (KSS) based on the correntropy maximization is proposed to nominate number of candidates from the space of aforementioned linear baseband MLD. In which, the higher order statistics (HOS) correntropy is exploited to compensate the mismatch between the MIMO-CEM received signal and the proposed linear approximation baseband MLD. Finally, the IF based MLD is applied on the nominated candidates to accurately estimate the transmitted sequence. The effectiveness of the proposed sparse MLD is tested and verified under different scenarios with different modulation techniques. On average, the proposed kernel sparse MLD achieves a complexity reduction by 99 % and 91 % compared to the conventional (IF-MLD) and recently MIMO-CEM decoder algorithm respectively.