Load Balancing Study Based On Node Ability

Ad Hoc network is the multi hop, self-organization and no center networkorganized by a set of mobile nodes with wireless transceiver. It is suitable for the situations of all kinds of no center communication node. However, if a single path and the traditional single channel routing protocol in Ad Hoc network are used, the network load is easy to focus on a small number of nodes, lead to node overload or congestion. At the same time, the channel competition between adjacent nodes will also affect other nodes on the same channel of communication. Therefore, Ad Hoc network needs the load balance technology to improve the existing deficiencies and improve the quality of communication.Through the system research on the load balance based on nodes ability in Ad Hoc network, the paper mainly achieves the effective early warning and the reasonable processing routing selection when a spurt of growth of network data. The study mainly includes three aspects as follow,(1) Studying on the shortest path, through comparing the three representative algorithm including the Dijkstra algorithm, Floyd algorithm and SPFA algorithm. At last, analyzing and summarying the advantages and disadvantages of them.(2)Through analyzing the characteristics of the wireless nodes resource usage, using information entropy to fit the node weight function closer to actual usage, and data modeling and topology structuring of adjacency set express. Comparing with the current node resource share in order to find the warning threshold through the node fitting function within the scope of upper limit warning thresholds M and lower limit warning threshold N. Warning in time if reacheing the warning threshold of a spurt of growth for better avoid.(3) In Guangdong radio and TV center machine room,using software of MRTG monitoring network link traffic load to collect and analyze the traffic information and utilization rating based on the distribution time on a certain amount of export route equipment, which has 200000 user data and analyzes the real and effective data. In the situation of static load balancing, using the using the Beta distribution to calculate the M and N of warning upper and lower limit warning threshold. Then, obtain the most realistic expectations of early warning threshold, make the routing switch when reaches the warning threshold to the realization of the aim of load balancing. Under dynamic load situation, combining the Dijkstra algorithm and the node weights of fitting function, process the collected data through mathematics, probability and sampling fitting. Then, as the data input model for early warning threshold entropy, by comparing the entropy value to make the route choice, so the node routing can be adjusted in the network, finally achieve the purpose of the node capacity load balancing, effectively avoid node upload and improve the quality of the network data transmission.