Research On Sampling To Generate And Data Optimization In The Participatory Sensing Networks

In order to improve sampling problems, data fusion , data processing problems and data serving problems in Participatory Sensing Networks, some algorithms are carried out in this thesis. The main contributions of this work are as follows:The first part of this paper designed a new algorithm based on the Dijkstra shortest path algorithm with the purpose of solving the problem that random sampling algorithm sample networks. First, using the Dijkstra algorithm samples the shortest path between nodes in social networks. Then, the frequency of edges in the path is ordered, and the edge of the higher frequency is selected. The algorithm solves some problems in the random sampling algorithm, and achieves a good function of the social network. Through the simulation experiment, it is proved that the sampling algorithm can better reflect the original network than random sampling methods.In order to improve the precision and reduce the complexity of traditional Multi-sensor data fusion,this part designed a two-step method using two algorithms.First, using FCM clustering algorithm analyses the collected data to eliminate the data which belongs to biased class. Second, using data fusion algorithm reduce data redundancy and get reliable data. Through simulated experiments, the results show that the protocol can make the data closer to the true value than traditional one step methods. It can realize the optimization of information in the Networks.In order to deal with the problem that data sparsity leads to the poor recommendation quality in collaborative filtering recommendation system. This paper presents a new collaborative filtering algorithm based on shadowed sets rough fuzzy c-means clustering. This algorithm solves the problem that nearest neighbor selection error due to the rating data sparsity. This algorithm also realizes the selection of the optimal neighbor sets of the collaborative filtering recommendation system. Compared with some traditional recommendation algorithms, it is proved that the proposed method can carry out the selection of neighbor sets better and can improve the quality of recommendation system better.All of the work has partly solved the sampling and data service problem in Participatory Sensing Networks, therefore could effectively analyze network ,improve service and save resource.