Source Controlled Modulation Scheme for Sources With Memory

Given an AWGN channel, the authors look at the problem of designing a source controlled binary antipodal signaling system for transmitting blocks of binary symbols generated either by a Markov Chain source (MC) or by a HiddenMarkovModel source (HMM). The goal is to minimize the average SNR required for any given block error rate. The particular case where the binary source is memoryless with nonuniform symbol probabilities has been studied by optimally allocating the signal energies. In this paper, they extend the previous work to include the important case of sources with memory.