Routing Algorithm Research Based On Node State In Opportunistic Social Networks

With the popularization of various intelligent devices and the development of various communication technologies,research on opportunistic networks has gradually become a hot topic in the network field.Researchers found that some theories of social networks can be transferred to opportunistic networks,and then form opportunistic social network with both characteristics.Opportunistic social networks still use the "store carry forward" mode in opportunistic networks to complete message transmission.Because it does not have the stable topology and mobile route in traditional network,how to use the social attributes of nodes to select appropriate nodes to complete data forwarding in opportunistic network environment is very important.At the same time,using link prediction method to select nodes can effectively increase the success rate of message delivery.However,link prediction schemes in opportunistic networks often only consider a single similarity index,while ignoring the complementarity between multiple similarity indexes.In addition,the design of caching strategy lacks the analysis of the encounter between the message and the destination node in the time to live,which will lead to the cache leaving some useless messages and wasting the node’s limited cache resources.Facing the above problems,this thesis has made the following contributions:(1)To solve the problem that the message attribute and node social attribute are not effectively combined in the traditional network,this thesis proposes a message propagation model based on message attribute and node state,which provides a basis for the selection of next-hop nodes in the process of message forwarding.Firstly,the message is classified based on the content,and then the message forwarding order is sorted according to the message category and node state.The transformation process between node states is simulated by the Markov chain,and finally the problem of message transmission selection of nodes in different states is solved,so that they can transmit the correct message to the destination node in the shortest time.Simulation results show that the model can improve the delivery success rate and reduce the average delay and overhead rate.(2)To solve the accuracy of link prediction in the network,and to solve the problem that the caching algorithm in the network rarely considers the encounter between the message and the target node in the time to live,this thesis proposes an inert cache replacement model based on historical prediction.The model uses three similarity indexes to predict the link,which reduces the instability of a single index.The performance of some critical nodes will decline sharply when they are dependent on the past.In view of this phenomenon,this thesis defines some nodes as inert nodes according to their low mobility and their historical information data.In the process of message forwarding,the message is distributed to inert nodes to avoid the death of some nodes due to high energy consumption.In addition,this thesis designs a cache replacement strategy based on node encounter probability and cache value,to make node cache more efficient.Experiments show that this strategy can effectively improve the message delivery rate and reduce the message delay.