Research On Heterogeneous Wireless Network Selection Strategy Based On Multiple Attribute Decision Making

With the continuous development of communication technology,different types of wireless network appears.Users are increasingly relying on the wireless network.How to make full use of the resources of different networks and provide users with the most satisfying quality of service in today's network environment coexisting with different types of wireless networks is an important problem to be solved in heterogeneous wireless network resource management.In the coexistence of a variety of different network resources environment,how to choose the most appropriate network for the users to access is the first question considered.In this paper,a characteristic utility function based adjusted cosine similarity network selection algorithm(CUFACS)is proposed.This algorithm uses adjusted cosine similarity to calculate the similarity between candidate wireless network attributes and the ideal network attributes,thus avoiding local convergence caused by large value local data in the main steam TOPSIS selection algorithms.Meanwhile,the differentiated utility function is introduced to normalize the attributes of different wireless networks.The influence of different attributes on the performance of network system and users' satisfaction is taken into account in the process of network selection,which makes the result of network selection more suitable for users.The experimental results show that the proposed algorithm can choose the best candidate network,and it will not fall into local convergence when the local data is too large.At the same time,the selection result ensures good network system performance and customer service satisfaction,and is more suitable in practical application.In order to continually providing users of the best network quality of services,allowing users to have the optimal network conditions,When the user's need or wireless resources change,the problem of network switching may occur,Users' seamless experience of the network switching is one of the key points of network service quality.Reduce the delay of network switching is the inevitable requirement for seamless switching network experience.In this paper,a BP neural network regression prediction algorithm based on L-M Bayesian regularization is proposed to predict the received signal strength of the next time slot terminal,and to determine whether the received signal strength reaches the network switching threshold.If the threshold is reached,the CUFACS network selection algorithm starts.Experiments show that the prediction error of this algorithm is small,and can make accurate predictions of the received signal strength in the next time slot,thus effectively reduce the network switching delay.