Research Of ACA And Its Application On Network Routing

Ant colony algorithm (ACA) is a new kind of meta-heuristic algorithms for solving combinatorial optimization problems or continuous function optimization problems, and it borrows the mechanism of ant colony's forage, and every ant is looked for as an agent. The characteristic of parallel, positive feedback and robust is strongly showed during ant colony's forage, ACA also has these characteristics, so the research of theory and utility of ACA has great merit.This paper analyzes the basic theory and model of ACA; introduces the features of ACA and its research state; selects some key parameters' value of ACA by simulations. The slow convergence and stagnation behavior are the main drawback of basic ACA, so we propose some methods to overcome these shortcomings. First, adopts the determinative and stochastic searching method to make selection; second, proposes the idea of subarea's searching to produce different quantity of pheromone in every route, accelerates the convergence of algorithm quickly in initial phase; third, combines with local and global pheromone's adjustment to update route's pheromone dynamically; fourth, alter some correlative parameters of pheromone adaptively; last, the idea of mutation is used to extend searching space. Puts forward a new ACA based on grid division to overcome the weakness of ACA in solving continuous function optimization problem, and expands the discrete space to the continuous space. For the research of application, analyzes the network routing optimization, studies optimization of QoS routing based on ACA in IP network; some other applications used by ACA are also been discussed.Simulations are made on some examples for the improved ACA proposed. Results of simulations demonstrate that the improved ACA is effective and feasible; simulations are also made for continuous function optimization and network routing optimization based on ACA etc. This paper could promote the research of theory and application of ACA.