Research And Implementation Of Adaptive Opportunistic Routing Algorithm Based On Machine Learning

With the advancement of wireless technology,various intelligent mobile devices have been widely used.Opportunistic network has become a research hotspot in the field of communication,because it does not require the establishment of a complete communication path between end to end,but pays more attention to the chance of encounter brought by node movement.Because of the lack of a stable end-to-end message forwarding path between nodes,how to deliver messages to the destination node efficiently and with low delay has become one of the hot issues.With the exploration of scholars,machine learning algorithms have made great progress in studying optimization problems.Therefore,this article combines machine learning methods to optimize and improve opportunistic routing algorithms to better perform adaptive message forwarding.The main work is as follows:(1)Aiming at the limitations of the current community routing algorithm considering only a single community,this paper proposes an Opportunistic Routing Algorithm Based on Clustering and Multi-community(ORACM).The algorithm first combines the advantages of DBSCAN and K-means algorithms to construct a hotspot location estimation model to calculate the hotspot locations of nodes,and clusters the hotspot locations to obtain clusters.Nodes may belong to multiple communities because they have multiple hotspot locations.In the process of message transmission,the ORACM algorithm makes full use of the multi-community characteristics and centrality of nodes for routing selection,and at the same time sets a redundant copy removal mechanism to reduce the invalid transmission of messages and avoid waste of resources.Finally,the real trajectory data is used to verify the ORACM algorithm.Compared with the existing classic routing algorithms,the ORACM algorithm has a higher delivery rate,a smaller network load and better performance.(2)Most of the current opportunistic routing algorithms only consider the intimacy between nodes,but do not consider their own attributes.This paper proposes an Opportunistic Routing Algorithm Based on Spectral Clustering and C4.5 algorithm(ORASCAC).The algorithm uses the SC algorithm to divide the community based on the intimacy between nodes.Secondly,the C4.5 algorithm is used to train the "messenger node" recognition model based on its own attributes.When delivering a message,if the source node and the destination node are in the same community,the centrality of the node is used to determine whether the message is forwarded.Otherwise,a "messenger node" is determined based on the trained model to deliver the message.Finally,the experiment shows that compared with the existing routing algorithms,ORASCAC algorithm has higher delivery rate,smaller network load.