Error Patterns Measured On Transequatorial HF Communication Links

As part of the MITRE Corporation's program for improvement of HF communication, the performance of coding and a means for evaluating coding equipment has been presented. It is, however, preferable to have a channel model upon which various error control techniques can be tried out as opposed to the direct approach, which is available only to the possessor of the data. This paper has been developed with the intention of presenting both the traditional channel modeling statistics (consecutive error distributions and gap distributions) and a set of new statistics against which modeling can be performed, The error statistics incurred in the transmission of high-frequency digital data at rates of from 600 to 2400 bit/s over various paths of the Eastern Test Range are presented herein. The data was transmitted on and parallel to the range using various modulation systems (modems). It is demonstrated that the errors occur in nonrandom fashion and in some cases are periodic. These error patterns (especially the periodic ones) present a new important problem for channel modeling of HF data From a heavily-used transequatorial circuit. While the basic approach is not new, the characteristics of the data are different from those previously reported and imply that channel performance is relatively independent of modulation technique. The error statistics presented herein should assist those, not having direct access to the data, who are interested in the problem of modeling this data which does not fit any known model (e.g. Gilbert, Berkovits, and Pareto[2]-[4]). Until such a model is available, this author will continue to report on error control performance using the direct simulation approach.