Network Threat Frequency Forecasting Based on Fuzzy Time Series and D-S Evidence Theory

Network threat frequency forecasting is the important factor for dynamic quantitative risk assessment, which can timely offer network security administrators the corresponding effective security strategy to actively counteract the network threat. This paper first reviews current time series forecasting methods, and then presents a network threat frequency forecasting algorithm based on fuzzy time series and D-S evidence theory. The proposed method first discusses the selection problem of interrelated forecasting factors, and then uses the selected factor-pair to construct factors order fuzzy logic relationships based on historical data to increase the forecasting accuracy. By experiments, the correctness and validity of the proposed method get validated.