Performance of a coded meteor burst system

The waiting time performance of a meteor burst (MB) communication system is investigated. The channel model used takes into account the time-varying SNR (signal-to-noise ratio) in an MB system. It is assumed that the message to be transmitted may be divided into several packets and that each packet may contain several code words from an (n, k, e) linear block code. The transmission protocol assumed is Protocol B of Miller and Milstein (see IEEE Trans. Commun., vol.COM-37, p.18-30, Jan. 1989), in which the transmitter has knowledge of when bursts are occurring and also uses a stop and wait ARQ (automatic repeat request) scheme. It is found that using a high-rate (≈0.85) block code can give improvement (in terms of average waiting time) on the order of 20%. The amount of improvement depends on the length of the message being sent to the assumed value of the decay constant τ. For small values of τ, an uncoded system performed almost as well as the coded system, whereas for large values of τ the coded system was significantly better