A Fault Prediction Algorithm Based on Rough Sets and Back Propagation Neural Network for Vehicular Networks

Vehicular networks (VNs) have become a feasible solution to solve network related problems in an intelligent transportation system. Due to their wide use in services, VNs are extremely vulnerable to interference that leads to frequent faults; therefore, their security challenges are widely recognized and urgently needed. Before a fault occurs, there are often some characteristic signals or data development trends that indicate a fault. If these data can be collected and judged effectively, fault prediction can be achieved. Therefore, it is of great significance to guarantee network security, reliability and continuous operation. In this paper, we propose a new fault prediction method for VNs. The fault prediction algorithm that uses rough sets and backpropagation (BP) neural network for VNs is divided into three parts, namely, the data acquisition module, the data prediction module, and the fault prediction module. The data acquisition module collects the data from the network to build the decision table using the rough sets and establishes the discernible matrix decision tables to reduce the data. The data prediction module combines the advantages of gray theory and BP neural network into a gray BP neural network to predict the data. The fault prediction module uses the normal data and fault data as input data and uses the fault type as the output to train the error BP neural network to obtain the appropriate weights, and then the predicted data is entered into the trained BP neural network to realize fault prediction. The fault prediction algorithm is simulated and analyzed using NS-2 and MATLAB, respectively, and the results show that the proposed algorithm can accurately diagnose and predict faults using the predicted data.