Bayesian Iterative Demodulation and Decoding for LoRa Modulation in Fast Rayleigh Fading Channels

Abstract

This paper designs Bayesian iterative detection and decoding (IDD) schemes incorporating a blind channel estimator in the physical layer of long-range wide-area network (LoRaWAN). The Bayesian IDD exchanges log-likelihood ratios (LLRs) as soft decision values between the LoRa demodulator and Hamming decoder to enhance the signal detection capability while obtaining the iterative gain. The data rate of LoRa is notably slow, consequently resulting in prolonged frame transmission times. In time-varying fast Rayleigh fading channels, the channel during the prolonged time experiences significant fluctuations due to the Doppler effect. Typical LoRa modulation uses non-coherent demodulation without channel estimation to overcome the Doppler problem, at the expense of detection accuracy. This paper investigates LLR for coherent demodulation, supported by an iterative blind channel estimator for tracking the time-varying channel states, in order to maximize the iterative gain in the IDD process. Finally, computer simulation results explicitly demonstrate the efficacy of the proposed Bayesian IDD in terms of bit error rate (BER) performance.