Outage prediction during intense rainstorm events using queuing theory and Markov Chains over radio links

Satellite communication links operating at higher frequency bands suffer from signal outages due to rain attenuation. Site diversity technique is one of the rain fade mitigation techniques that can be employed over earth-satellite links to improve on system availability. In this study, we use 5-year rainfall rate statistics and the queuing theory approach to investigate the attributes and behavior of intense rain storms along an earth-space link over Durban, South Africa (29^◦ 52’S, 30^◦ 58’E), a sub-tropical climate. Thereafter, a comparison is made with results obtained in a related study in Jimma, Ethiopia (7.6667^◦ N, 36.8333^◦ E), which is a tropical climatic region. Verification of the best fit distribution is done through the application of the root mean square error (RMSE) and CHI squared statistics. Results of these analysis tools confirm the suitability of the proposed distributions with RMSE error margin in the range 0.0024 to 0.0128, and a χ² statistics value of 0.4070. The spike service time for such rain storms is found to follow Erlang-k distribution in both regions of South Africa and Ethiopia as opposed to earlier determined exponential distribution. In addition, the analysis shows that there exists a power law relationship between the rain spike maximum rain rate and its diameter. This relationship is further utilized in the development of the rain cell sizing model that can be used for site diversity fade mitigation. Furthermore, the Markov chain technique is employed to determine the occurrence behavior of shower and storm rainfall regimes, and their contributions to rain attenuation over a slant path radio link.