Application Research Of Probabilistic Modeling Evolutionary Algorithm On Routing Problem Of Ad Hoc Networks

With the development of wireless ad hoc networks, the multipath routing and QoS routing become the key issues in recent years. Probabilistic Modeling Evolutionary Algorithms (PMEAs) are a new class of evolutionary algorithms by incorporating the building models into evolutionary algorithms. And Bayesian optimization algorithm (BOA) is a representative one among PMEAs in solving high order hard problems. In this paper, we mainly study the applications of BOA and multi-objective Bayesian optimization algorithm (MBOA) to the fields of multipath routing and QoS routing in wireless ad hoc networks. The main works in this paper are as follows:It has been recognized that the optimized multipath routing problem is NP-hard, which, in this paper, has been mapped into a problem that BOA can solve. Based on BOA's advantage of optimizing population, we adopt it to optimize multiple paths and find a set of optimum paths. Simulations indicate that BOA has quicker convergence, better performance of searching for the optimum solution from the space of solutions and getting lower distortion. Moreover, BOA can satisfy the real-time multimedia communication.Since the delay-constrained QoS routing is NP-complete question, BOA is applied to solve it in this paper. To find the optimized solution, one reasonable coding method and fitness function are introduced into BOA, and Bayesian network is used to simulate, model and sample the object parameters also. Experimental results show that BOA has better performance in optimizing delay-constrained QoS routing and can search lower cost routing with the least message overhead. Obviously, it meets the needs of QoS routing.In order to improve the performance of QoS routing in ad hoc networks, in this paper, the bandwidth is chosen as constraint condition. The delay-time, cost and lost-packet ratio are chosen as the optimized goals. Then, we build up the model of multi-objective optimization and use MBOA to determine the model. The theoretical analyses and experimental results show that MBOA can maintain the variety of solutions and search the better Pareto fronts with quicker...