Research And Implementation Of Clustering Algorithms For Broadband Micro-power Network Communication Systems

Broadband micro-power network adopts wireless transmission mode,which is mainly to provide users with higher transmission rate,wider coverage area and more stable network topology.The network layer of broadband micro power protocol stack is mainly responsible for networking and network maintenance.Although the time of network topology stability can be accelerated in the current networking strategy,it does not consider the constraints of proxy coordinator nodes and the hierarchical management of network nodes,which leads to poor network scalability,high end-to-end average delay,high probability of network congestion and so on.Therefore,through the research of clustering algorithm,this thesis proposes a clustering algorithm suitable for broadband micro-power networks.The algorithm is divided into cluster formation stage and cluster tree formation stage.(1)Cluster formation stage: Firstly,the characteristics of broadband micro-power network and five classical distributed clustering algorithms are analyzed in this thesis.Then,aiming at the problems of boundary nodes and node connectivity when EADEEG algorithm is applied to broadband micro-power networks,the algorithm is improved from two stages: node cluster head competition and neighbor non cluster head node selection.The improved algorithm is an algorithm based on node energy,the number of neighbor nodes and the link quality indication value of nodes.The simulation results show that under the same conditions,the number of cluster head nodes generated by the improved algorithm is less and evenly distributed than that generated by EADEEG algorithm,and the connectivity of cluster head nodes is guaranteed.(2)Cluster tree formation stage: A shortest path data aggregation cluster tree algorithm is proposed in this thesis.In the cluster tree establishment stage,the algorithm uses Fibonacci heap as the priority queue to optimize Dijkstra algorithm,and obtains the shortest path from cluster head node to central coordinator node through the shortest path algorithm.After the cluster tree is stable,the number of discovery list messages sent by nodes is reduced through data collection technology,and the data of member nodes is collected at the cluster head node to reduce the correlation of messages in the network.Through the performance analysis of the algorithm,the algorithm can effectively reduce the layer of PCO nodes in the network,reduce the length of message and decrease the occupation rate of channel resources.Finally,this thesis implements the clustering algorithm in the broadband micro power network layer,including simplifying the message and through the analysis of CSMA time slot.It is verified by building a broadband micro power network experimental platform.The verification results show that the clustering algorithm can effectively reduce the number of cluster head nodes in the network,decrease the layer of network nodes,improve the service execution efficiency,and prolong the number of meter reading rounds.