ANOMALY DETECTION OF NETWORK SWITCHES PERFORMANCE IN I&C SYSTEM BASED ON NETWORK CALCULUS

抄録

In this paper, a network switch performance anomaly detection method of the instrumentation and control (I&C) systems in nuclear power plants (NPP) based on network calculus is proposed. In the Ethernet-based packet switching I&C system network, the delay of a network switch is an important indicator of network performance. With the development of small modular reactor (SMR) technology, the scale of the I&C systems in NPPs has increased. The I&C systems of SMRs are often constructed in batches with a modular architecture, which imposes higher requirements for the extensibility of the network architecture and network performance. Traditional network performance analysis uses the queuing theory based on the statistic method and the stochastic process to analyze the average performance of the network, instead of analyzing the deterministic boundary of system performance. In this paper, anomaly detection methods based on network calculus are proposed for network switches in the NPP I&C system by detecting rate-latency performance, delay, and data backlog anomalies. Network calculus is a deterministic queuing theory based on nonlinear algebra. It takes the data flow as the research object, obtaining the performance bound of network nodes through the convolution and deconvolution of the min-plus algebra. The MATLAB simulation experiments are conducted and verify the effectiveness of the anomaly detection methods.