Centralized broadcast scheduling in packet radio networks via genetic-fix algorithms

An important, yet difficult, problem in the design of a packet radio network is the determination of a conflict-free broadcast schedule at a minimum cycle length. We first formulate the problem via a within-two-hop connectivity matrix and then, by assuming a known cycle length, determine a conflict-free scheduling pattern using a centralized approach that exploits the structure of the problem via a modified genetic algorithm. This algorithm, called genetic-fix, generates and manipulates individuals with fixed size (i.e., in binary representation, the number of ones is fixed) and therefore, can reduce the search space substantially. We also propose a method to find a reasonable cycle length and shorten it gradually to obtain a near-optimal one. Simulations on three benchmark problems showed that our approach could achieve 100% convergence to solutions with optimal cycle length within reasonable time.