Network Security Situation Prediction System Based on Neural Network and Big Data
Bowen Zhu, Yonghong Chen, Yiqiao Cai, Hui Tian, Tian Wang · International Journal of Security and Its Applications · 2017
In today's big data era, traditional methods are of low efficiency in handling network security matters, and most of the time they even don't work.The system studied in this paper, the network security situation analysis and prediction system based on neural network is designed and implemented on the Hadoop platform.By collecting distributed data and decreasing their dimensions, this system reduces the complexity of data to realize efficient processing of big data.We adopt the optimized K-Means clustering analysis algorithm to simplify the data, and we utilized the optimal association rules mining method to find threats and risks existing in the network.The above part is the network security situation analysis.On the basis of network security situation analysis, a new method based on time dimension is used to forecast the future network security situation.By blending part of predictive results and adjusting error, the system realizes security situation prediction of the whole network and a self-improving neural network, thus ensuring a higher accuracy rate.The experimental result we obtained is that the time we spend is just 12% of what consumed by the traditional method in the same amount of data.We can draw the following conclusions: 1) the system proposed in this paper can effectively save time of handling big data 2) as the amount of data increases, this system will not reduce the accuracy rate but gets 95% correct.