A comparison of protocols for a meteor-burst channel based on time-varying channel model

Abstract

The performance of a meteor-burst (MB) communication system using three different protocols is examined. The protocols were chosen to study the relative advantages of using a simple stop-and-wait automatic repeat request (ARQ) scheme as well as the ability to sense the presence of a channel. It was found that for long messages, the use of the ARQ scheme gave significant performance improvement; however, when the message length was on the order of the average burst length, the use of the ARQ scheme was not helpful. For packet sizes on the order of the average burst length, it was found that the ability to sense the presence of the channel gave about a 20% reduction in the time necessary to complete a message, but for much shorter packet sizes, the channel probing was not needed. The model used to analyze these systems takes into account the time-varying signal-to-noise ratio (SNR) present in a MB channel. The performance predicted by this model is compared to a less sophisticated constant SNR (signal-to-noise) model and its is found that the author's model predicted significantly more optimistic performance. >