The Research On Bayesian Network Congestion Control For Ad Hoc Networks

In wireless Ad Hoc networks, the node acts as both transceiver and router that it forwards data for other nodes. It makes that the Ad Hoc network is more prone to congestion, and network congestion is an important factor that restricts the performance of the network. Therefore, in order to make better use of the networks resources and avoid congestion occurring and ensure network quality of Service, a congestion control algorithm based on Bayesian Network is proposed in this thesis. In order to make better use of the real-time datum that achieved from the Ad Hoc network, the congestion control algorithm based on Bayesian Network is improved. Realization processes of the algorithms are as follows:Firstly, it is data processing. In order to make use of the data to create the Bayesian Network and use the achieved Bayesian Network to reason, it needs classify the datum. Classifying the data is by delimiting the threshold in the congestion control algorithm based on Bayesian Network for Ad Hoc networks. While it uses the fuzzy logic to handle the data in the improved method for the improved method for the congestion control algorithm based on Bayesian Network for Ad Hoc Network.Secondly, it constructs Bayesian Network. Firstly, the K2algorithm is used to learn a structure of Bayesian Network from that the processed datum, it can achieve the structure of Bayesian Network. Then the maximum likelihood function is used to learn the parameters of Bayesian Network on the achieved the structure of Bayesian Network, it can achieve Bayesian Network.Thirdly, it predicts status of the Ad Hoc networks. In order to ensure that the predict status of the Ad Hoc network is accurate, it reasons the probability of various states by the obtained real-time parameters of Ad Hoc network. The Bayesian Network inference is used to predict status of the Ad Hoc network in the congestion control algorithm based on Bayesian Network for Ad Hoc networks. But an improved Bayesian Network inference is proposed, which is used in the improved method.Fourthly, it implements the congestion control. In order to achieve the adaptive adjustment for the congestion in Ad Hoc network, it needs adjusted the sending rate of the network in time according to the achieved prediction results. The sending rate adjustment method based on the status of the Ad Hoc networks is designed in the congestion control algorithm based on Bayesian Network for Ad Hoc networks. But the sending rate adjustment method based on the probability of the status in multiple time periods is designed in the improved method. Finally, it simulates the algorithms which are proposed by this paper. The simulation results show that both algorithms to certain extent can avoid the congestion, and they can adaptively adjust the sending rate according the status of the Ad Hoc networks, and they ensure the network Quality of Service (QoS).