Imitating the human immune system capabilities for multi-agent federation formation

In this paper, we are trying to highlight specific properties of the immune system, in order to develop the immune optimization algorithm as an optimal solution for the multiagent federation formation problem. Behaviors of the antibodies as basic agents of the immune system are considered and their collaboration structure is studied to form a relevant algorithm for multi-agent control problems. In immune system the optimization problem is addressed by considering two functions: fitness and affinity. Fitness, which is the goal function for optimization, is exterior for the agents group, and affinity function is the internal factor among agents. Therefore the cost function is divided to the two independent parts. The second part is distributed among agents as affinity function. We compared our proposed method with the regular genetic algorithm and show some simulation results on multirobots federation formation.