A Link Prediction Method for Opportunistic Networks

Abstract

The invention discloses a link prediction method for an opportunity network. At first, all possible nodes are classified and divided into node pair sets meeting periodically, node pair sets meeting frequently in a non-periodical mode, and node pair sets meeting infrequently; then different methods are used for carrying out link prediction on the different node pair sets, a periodic pattern mining method is used for predicting the node pair sets meeting periodically, a J48 decision-making tree method is used for predicting the node pair sets meeting frequently in a non-periodical mode, and an Adamic-Adar algorithm in a complex network is used for predicting the node pair sets meeting infrequently. By means of the mode, the problem that the application range of a single link predicting method is limited is solved, the precision of link predicting of the opportunity network can be improved, the recall rate of link predicting of the opportunity network can be increased, the message delivery efficiency of the opportunity network is improved, and the capacity of the opportunity network is increased.

Description

technical field [0001] The invention relates to the field of network technology, in particular to a link prediction method for opportunistic networks. Background technique [0002] 1. The development history of the technical field [0003] Opportunistic network is a new type of network architecture developed from mobile ad-hoc network. It can use the encounter opportunities brought by node movement under the condition of segmented network to realize the hop-by-hop forwarding of packets and finally deliver them to the destination node. The concept of encounter refers to a connection that occurs between nodes: when nodes enter the communication range of each other (such as within the point-to-point communication distance of Wi-Fi or Bluetooth protocol), a link is established and communication occurs, and when the two leave When they are within the communication range of each other, the link is disconnected and the communication is stopped. Due to the mobility of nodes, encou...

AI Technical Summary

Problems solved by technology

There are still some problems in the existing opportunistic network: (1) The success rate of message delivery is not high, and the message will be discarded during the delivery process if it exceeds the lifetime or the buffer overflows. The delivery success rate is generally 50%; (2) The message delivery The delay is high, and due to the blindness of the storage due to the dynamic nature of the network, it may take hours or days to successfully deliver the message; (3) multiple copies are stored and forwarded in the network, which will inevitably occupy a large amount of storage and Lead to high energy consumption; (4) Due to the lack of real-time network, whether the network service is reachable or not, the service waiting time is also unpredictable, which significantly reduces the user experience

Obviously, the prediction of future encounters between periodic node pairs, non-periodic node pairs, and node pairs that have never met before is a different problem. The existing single prediction method cannot handle all prediction problems. It does not discuss how to divide different types of node pair sets, and how to choose the optimal prediction method on each node pair set

Images