Adaptive Binary Coding Schemes for Trading Data Rate for Reduced Signal to Noise Ratio Requirement.

摘要

A detailed analysis is made of methods of increasing the redundancy of binary level data words so as to permit transmission at a reduced signal to noise ratio without an increase in error probability. Repeated transmission of the same data word is shown to be rather inefficient, especially if the integration is carred out after stages of non-linear detection and/or decoding. The need for care in ensuring the statistical independence between replicates, and between the individual bits of longer codewords, is shown. A cascade of a Golay cyclical error correcting coding process, followed by an orthogonal sparse coding, using maximum amplitude detection instead of threshold detection, is found to provide an efficient system where signal to noise ratio is restricted and energy consumption at the transmitter must be minimised. Replication can be used to provide an additional trade off during periods of abnormally low signal to noise ratio. An analysis is also given to enable the calculation of losses arising from commonly used quasi-matched filters for the extraction of signal elements from Gaussian noise. (Author)