Binary shaping for low-duty-cycle communications: a useful alternative to PPM

We develop a coding method called binary shaping that generates a binary channel input of arbitrary duty-cycle with Bernoulli statistics, which is useful for laser communications using generalized on-off keying. We couple binary shaping with a serially concatenated turbo code for forward error correction, and show by simulation that the performance of this scheme with iterative decoding approaches the capacity of generalized on-off keying on the Poisson channel for medium duty cycles, significantly exceeding the performance of pulse-position modulation (PPM) in this regime. Analogous to trellis shaping for the AWGN channel, binary trellis shaping utilizes a convolutional shaping code to which we apply the Viterbi algorithm at the encoder to minimize Hamming weight. We show how low-duty-cycle communications is a special case of information embedding and discuss how binary shaping relates to information embedding methods.