Iterative decoding techniques for ring-type magnitude modulated signals

The expectations for fifth generation (5G) systems involve handling huge throughputs together with power consumption. In order to address this issue, massive multiple-input and multiple-output (MIMO) schemes were developed exploring the fact that any M-ary constellation can be decomposed as several offset quadrature phase shift keying (OQPSK) signals, to be amplified by efficient saturated amplifiers and transmitted separately. Since nonlinear amplifiers should be employed with constant envelope signals to avoid nonlinear distortion, a ring-type magnitude modulation (RMM) technique was proposed for bandlimited OQPSK signals to mitigate its envelope fluctuations while maintaining its spectral characteristics. This technique was shown to improve the power amplification efficiency of linear amplification with nonlinear components (LINC) transmitters, although with some degradation of the of the detection performance. This paper proposes an iterative decoding scheme relying on the accuracy of the first iteration's log likelihood ratio (LLR) computation to estimate the RMM coefficients of the received symbols. Several LLR computation methods are compared with the ideal case where the receiver knows each symbol's RMM coefficients, and it is shown that a few iterations are enough to obtain a bit error rate (BER) performance close to the best case scenario.