Opportunistic Network Performance Optimization Model Based on a Combination of Neural Networks and Orthogonal Experiments

Abstract

Under the complex environmental conditions of remote geographic locations and fragmented network areas, sending data over traditional wired networks or wireless self. Organizing networks often results in connection interruptions, packet loss and other problems. In order to reduce the problem of data loss due to network disruptions, researchers have proposed a new network technology for cut networks, named opportunity Network. Opportunity network messages have a high rate of packet loss due to factors such as the small probability of node encounters and message lifetime limits. In order to reduce the packet loss rate, improve the success rate of message transmission and reduce the message transmission delay, the paper proposes a node message transmission performance optimization model based on the fusion of neural networks and orthogonal experiments. The experimental results show that the prediction model proposed in the paper for optimizing node message transmission performance can predict the packet loss rate more accurately, and it is found that node density, node rate, and node cache are the most important factors influencing the message packet loss rate of the opportunity network, and the optimal settings of these three factors are obtained to reduce the packet loss rate of message transmission and improve the message delivery efficiency.