Research On High Throughput Channel Allocation Technology In Wireless Sensor Networks

Wireless sensor network(Wireless sensor network,WSN)is a kind of distributed sensor network,the Wireless sensor network in the mainstream of the research direction is the channel assignment method,with the development of information technology,sensor network users in the growing and network topology is also becoming more and more complex,and the scarce Wireless resources and the contradiction between the high density of AP point will have bad for People's Daily life,in order to better solve this contradiction,dynamic channel allocation method is one of the important methods to deal with the practical problem.Wireless sensor network(Wireless sensor network,WSN)is a kind of distributed sensor network,the Wireless sensor network in the mainstream of the research direction is the channel assignment method,with the development of information technology,sensor network users in the growing and network topology is also becoming more and more complex,and the scarce Wireless resources and the contradiction between the high density of AP point will have bad for People's Daily life,in order to better solve this contradiction,dynamic channel allocation method is one of the important methods to deal with the practical problem.Many of the methods for the current wireless channel allocation,this paper designed based on the heuristic algorithm,and then by using the method of uniform distribution to optimize allocation,evolution of the principle of continuous innovation,through a genetic algorithm with multiple iterations updates,produce multiple sets of allocation strategy as a machine learning of the original data set.Then by comparing the topological similarity,the new input topology is pre-allocated,and the throughput obtained by calculating the pre-allocation scheme is compared with the optimal throughput obtained by using the idea of genetic algorithm to allocate the new topology,and the merits of the new algorithm are judged.The network model is constructed by means of scatter map,density map and thermal map,and the optimal network model construction method is selected bycomparing the differentiation degree of different network models.In order to verify the performance of the algorithm,based on the spyder analog wireless sensor network,a small region associated with hot to high density AP point topology information,using heuristic algorithm and recommended in this paper,the improved genetic algorithm and the algorithm simulation strategy distribution respectively,with total throughput of network as the metrics and compare the advantages and disadvantages of three kinds of allocation strategies,which proved that the presented algorithm is advanced.Because the network topology information and the allocation strategy in the database are one-to-one correspondence,the search quantity is very large,so machine learning can not be used to label.In order to make the allocation strategy more universal,we can use machine learning to label and cluster the allocation strategy by clustering.The machine learning model is trained by Banach theorem,and the model feasibility is proved by simulation.The machine learning model is applied to the network in this paper,and the algorithm is optimized by gradient descent rule.