Algorithms For Edge Cloud Clustering Based On Maximal Clique

Mobile-edge cloud computing is an emerging paradigm,which provides localized cloud computing in the edge of the network for applications such as car networking,smart wearing,virtual reality,and augmented reality.It can reduce the communication delay between users and the server.Coordinating multiple edge clouds is a promising focus for improving the performance of edge clouds.The delay between nodes restricts the performance of edge clouds.Hence,the design of an optimal node combination with lowest delay is significant.This paper mainly researches on the edge cloud clustering algorithms to minimize the global maximum delay between edge cloud combinations.Firstly,a topology graph is modeled based on edge.An edge cloud node is represented as vertex.The weight of the vertex is the idle resource,such as the number of CPU.The edges between vertexes represent the connections and the weight of edge is the delay between corresponding cloud nodes.To overcome the deficiency of min Star with selecting of incomplete graph,which increases the congestion of network and the communication delay between nodes,the edge cloud clustering algorithm based on maximal clique is proposed.The proposed algorithm divides the topology into a number of non-overlapping complete sub-graphs,then encapsulates the sub-graph into a resource pool.The resource pool contains the basic information of the composition corresponding to the sub-graph.The resource pool is selected,which satisfies the user request with the minimum of communication delay.The proposed algorithm combines the nodes with a group to ensure the direct communication for any two nodes.In addition,we design a task migration algorithm for edge cloud combination,which illustrates the flexibility of the new algorithm.Experimental results demonstrate that the proposed algorithm can reduce the global maximum communication delay to half of the cloud combination in the system when the utilization of the system resources is up to 2/3.