Since well-known decoding algorithms [1],[2] are able to correct both character errors and character erasures withq-ary Reed-Solomon codes, some modulation and demodulation schemes designed to be used with such codes provide a channel that hasqinputs andq + 1outputs,qof which correspond to theqinputs and one of which corresponds to the "erasure" symbol. This correspondence points om the relative advantages offered by a slightly more refined demodulation scheme, which creates a digital channel that hasqinputs and2_{q}outputs,qof which correspond to "strong" receptions of the inputs andqof which correspond to "weak" receptions of the inputs. For decoding purposes, allqweak outputs may be treated as erasures, but the fact that the channel now provides additional information facilitates improved decoding performance by reading the weak characters under the erasures.

Four realizations of interleavers that reorder a sequence of symbols so that no contiguous sequence ofn_2symbols in the reordered sequence contains any pair of symbols that were separated by fewer thann_1, symbols in the original ordering are introduced. For anyn_1andn_2that satisfy an appropriate relative primeness condition, these interleavers are optimum in the sense that of all possible interleavers providing the indicated symbol separation, one of these four realizations achieves both the minimum possible encoding delay and the minimum possible combined storage capacity for the interleaver and its unscrambler.

The suitability of using error-correction coding to improve the performance of an HF AUTODIN circuit is investigated. It is demonstrated that coding is not a practical method of improving the performance of HF AUTODIN circuits that possess good frequency complements and can make effective use of them.