Spectra of multimode coded signals

Multimode coding is an efficient block coding technique that introduces control over sequence statistics by generating a number of alternatives to represent the source word in each encoding interval and then selecting the word that best meets the system constraints. The power spectral density (PSD) of the encoded signal is of particular interest with these codes. Standard techniques for evaluation of the PSD of block coded sequences cannot be directly applied to a wide variety of multimode codes in which encoder state probabilities do not reach a stationary distribution, or in which codeword selection is random when two or more alternatives satisfy system constraints. In the paper standard spectral analysis techniques are extended to enable evaluation of the PSD of signals generated by multimode codes with these characteristics. It is demonstrated that randomness in selection can result in suppression of discrete components in the encoded signal that may otherwise arise under adverse conditions.