Capacity and Nonuniform Signaling for Discrete-Time Poisson Channels

The Poisson photon-counting model is accurate for optical channels with low received intensity, such as long-range intersatellite optical wireless links. This work considers the computation of the channel capacity and the design of capacity-approaching, nonuniform signaling for discrete-time Poisson channels in the presence of dark current and underaverage and peak amplitude constraints. Although the capacity of this channel is unknown, numerical computation of the channel capacity is implemented using a particle method. A nonuniform mapper is coupled to a low-density parity check code and a joint demapper-decoder is designed based on the sum-product algorithm. Simulations indicate near-capacity performance of the proposed coding system and significant gains over information rates using traditional uniform signaling. A key observation of this work is that significant gains in rate can be achieved for the same average power consumption by using optical transceivers with nonuniform signaling and a modest increase in peak power.