Improving bacterial foraging algorithm using non-uniform elimination-dispersal probability distribution

Bacterial foraging optimization (BFO) is a swarm intelligence algorithm inspired from forging behavior of the E. coli bacteria. The BFO is based on three basic processes; chemotaxis, reproduction, and elimination-dispersal. In the elimination-dispersal process, a constant probability of elimination is assigned to all bacteria. The assignment is independent of bacteria’s ranking in the population. Therefore, a bacterium which can be near to an optimal position may be replaced with one which is far away from the optimum solution, thus, affecting the convergence speed. In this paper an improved BFO algorithm (IBFOA) is proposed. The improvement lies in the elimination and dispersal event in which a non-uniform probability distribution is used. The non-uniform distribution is implemented by a linear and nonlinear probability distribution that replaces the conventional constant distribution. Simulation results show the efficiency of the proposed algorithm compared to the conventional bacterial foraging algorithm in terms of convergence speed. Keywords: Bacterial foraging algorithm, Metaheuristic algorithms, Elimination-dispersal event, Non-uniform probability distributio