Soft-metric based decoding for photon counting receivers

In [1] a low-complexity photon-counting receiver has been presented, which may be employed in long-distance amplification-free classical optical communication schemes, and which is typically modeled through its equivalent Binary Symmetric Channel (BSC) model. In this paper, we consider the scheme described in [1], but we model it as a time varying Binary Input-Multiple Output (BIMO) channel, and analyze its performance in presence of soft-metric based capacity approaching error correcting codes. We show that the classical channel capacity of the suggested BIMO model is higher than the capacity of the BSC model, and that the use of the BIMO model allows to feed the channel decoder with soft information, in the form of Log-Likelihood Ratios (LLRs), achieving a significant reduction in Bit Error Rate (BER) and Frame Error Rate (FER) with respect to classical hard-metric-based schemes which should be used in conjunction with a BSC channel model.