A Novel Weighted LLR Algorithm for Non-Coherent OFDM-MFSK System in Ionospheric Scatter Channel

A novel weighted log likelihood ratio (LLR) algorithm is proposed for the non-coherent orthogonal frequency division multiplexing with
${M}$
-ary frequency shift keying (OFDM-MFSK) system in an ionospheric scatter channel (ISC) with serious frequency-selective fading. In the algorithm, the bit LLR is derived according to the probability distribution of non-coherent received signals in the MFSK symbol and is approximately simplified to a concise expression, which reveals that LLRs cannot indicate the effect of the frequency-selective fading channel on different MFSK symbols. The proposed weighted LLR algorithm maximizes the signal-to-noise ratio (SNR) of the metric combined with LLRs in channel decoding. The weighted factors were estimated blindly from the received MFSK signals. Simulation results show that the bit error rate (BER) of OFDM-MFSK systems with the proposed LLR-weighted algorithm is superior than that of conventional LLR algorithms, and has approximately 1.6 dB SNR gain when BER is
${10^{-4}}$
.