Intrusion Event Identification for Fiber Perimeter Security System Based on ARMA Modeling and Sigmoid Fitting

Intelligent intrusion detection for optical fiber perimeter security system based on an improved high efficiency feature extraction technique

“…Table 3 gives the comparison results between the proposed scheme and the reported schemes in [16,17] and [19]. We can see from table 3 that compared with the reported schemes for identifying the six given intrusions, the proposed intelligent detection scheme exhibits superior performance in terms of both processing time and recognition rate.…”

“…And to further verify the efficacy of our proposed scheme that can accurately recognize normal environmental interference events, we also collected sensing data of natural environmental interference using the DMZI-based OFS system. [16] 92.68% 0.32 s Huang et al [17] 84.17% 1.06 s Lyu et al [19] 84.76% 0.17 s Figure 12 shows the time domain waveform of normal environmental interference and their corresponding kurtosis vectors. It can be observed from figure 12(a) that, compared to the time domain waveforms of intrusion events, the amplitude of normal environmental interference changes slowly and randomly.…”

“…For example, in 2018, Ma et al proposed an intrusion event recognition scheme based on a multiscale permutation entropy based-feature extraction method and a probabilistic support vector machine based-classifier [16]. This scheme can distinguish six types of intrusion events with a mean recognition rate of 92.68% and a mean processing time of 0.32 s. In 2020, Huang et al developed an intrusion event scheme combining an autoregressive moving average (ARMA) model and a support vector machine classifier [17]. In particular, the ARMA coefficients are used as the features to classify the unknown intrusions.…”