An Improved Ant Algorithm for Network Traffic Control

This paper proposed an improved ant algorithm with feedback function extension and dynamic pheromone design (dynamicAnt) for the network traffic management issue. The scheme first mapped the network traffic path delay and bandwidth metrics into the parameters of the basic ant algorithm. And then extended the network feedback function to the basic ant algorithm by simulated it as the food smell. By releasing the food smell and the ant pheromone with a dynamic design, the path selection behavior of each ant will be influenced according to their variation. The aim of dynamicAnt lied in affecting the ants´ behavior to make fully use of available paths with avoiding the paths congestion in the network traffic control. Besides analyzing the effective control of the pheromone and its related transfer possibility, compared with basic ant algorithm, simulation experiments illustrate that the proposed algorithm achieved better network traffic performance in network throughput, path delay and packet loss rate.