Network Selection Algorithm Based On Personalized Requirements And Behavior Characteristics Of Users

Wireless communication technology is developing rapidly.Wireless communication technology has developed into a heterogeneous wireless network with different bandwidth,modulation mode and coverage.Heterogeneous wireless networks provide users with seamless roaming and transparent services everywhere.Academics have proposed various algorithms for network selection in heterogeneous wireless networks.But most of these algorithms do not make effective use of historical data.Two algorithms that make full use of historical data are proposed to mine the individual requirements and behavior characteristics of users,thus optimizing the performance indicators of network selection.The main works of this thesis are as follows:1.In the heterogeneous wireless network,a network selection algorithm which can greatly improve the satisfaction of users is proposed based on the analysis of historical data to deeply mine the personalized requirements of users.First of all,historical access data and current candidate network data are obtained.The parameters are classified into three categories: cost,system performance and stability according to individual needs.The expected and current benefits of the three categories of attributes are calculated using the fuzzy analytic hierarchy process.The psychological curve function is introduced based on benefits mentioned above to build the user satisfaction model,which represents the user's satisfaction degree to each attribute demand of the candidate network.Secondly,the radar graph analysis method based on growth curve function optimization is introduced to calculate the comprehensive satisfaction of candidate networks and select the network access with the best satisfaction.Finally,three experimental simulations show that this algorithm can improve user satisfaction,meet user's personalized needs,and effectively reduce the ping-pong effect while ensuring load balance.2.In the ultra-dense heterogeneous wireless network,a network selection algorithm which can greatly reduce the switching delay and resource consumption is proposed based on the analysis of historical data to mine behavior characteristic information of users.First of all,the offline data analysis is carried out through the historical behavior of each user.And the collaborative filtering recommendation algorithm is used to get the user similarity evaluation and give the recommendation network.Then select the recommended network for access.If the recommended network is not available,raccoon optimization neural network algorithm is used for network selection.Secondly,the comprehensive score of each candidate network and the user experience score of access network are updated in real time.An online feedback and scoring mechanism is constructed to update the historical database,so as to ensure the accuracy of the recommendation algorithm.Finally,three experimental simulations show that this algorithm can effectively reduce switching delay and signaling overhead.And because the online feedback scoring mechanism takes into account metrics such as blocking rate,users will avoid networks with higher blocking rates when switching decisions,thereby reducing the average blocking rate.