Performance comparison of quasi-static routing algorithms for packet-switched computer networks
Nabil Ayad, Fihri Mohammed, Loutfy H. Madkour, M. S. Metwally · ACM SIGCOMM Computer Communication Review · 1985
The routing problem is one of the most important problems facing the development, improvement, and performance of packet switched computer networks (PSN). In this paper two, quasistatic techniques, which proved to be robust with respect to overall performance, are compared. The first one, uses "The Learning Automata" principle, which is a promising technique regarding its simplisity and ease of implementation. This technique is thoroughly investigated. The second, uses "The Deterministic Routing" principle, that is selected for comparison since it uses the deterministic sequence generation like the first one. The goal is to highlight the efficiency of each technique with respect to the other. This have been realized using a simulated 10-nodes sample network to clarify the advantages and defficiencies of each technique. The relative performance of the two algorithms is judged on the basis of delay and blocking probability. The results are in favour of the learning automata specially in cases of networks of big number of nodes.