Design of routing tables for a survivable military communications network using genetic algorithms

One of the vital areas in the design and operation of a survivable military telecommunications network is the selection of its routing tables. Both a bit-string genetic algorithm (GA) and an iterative stochastic hill climber (ISHC) are applied to two problem scenarios: with and without existing routing tables. Experimental results are reported for one destination node in an 18-node model network. Comparisons are made both between the GA and ISHC approaches, and with the preexisting routing table. Overall, both the GA and the ISHC provide substantial improvements over the existing routing table, for both problem scenarios. However, the ISHC consistently obtains better results than the GA, often at reduced computational cost.