论文部分内容阅读
针对传感器网络中由于传感器故障造成的异常点检测问题,该文提出一种基于传感器与其空-时近邻点在测量数据之间的差异,采用似然比检验来判断传感器是否故障的异常点检测方法。在空间维,该方法基于最大后验概率选取待检测传感器当前时刻的空间近邻点;在时间维,该方法选取待检测传感器在之前若干个时刻的测量值作为其时间近邻点。然后根据待检传感器与其空-时近邻点测量数据之间的差异对其异常程度进行量化,并采用似然比检验判断待检测传感器是否故障。结果表明:该方法与已有的异常点检测方法相比,在相同的虚警率下取得了更高的检测率。例如在虚警率为10%时,该方法将检测率提升了10%~30%。
Aiming at the problem of abnormal point detection caused by sensor failure in sensor networks, this paper proposes a method based on the likelihood ratio test to determine whether the sensor is faulty based on the difference between the sensor and its space-time neighbors in the measured data . In the space dimension, the method selects the nearest neighbor of the sensor to be detected based on the maximum posteriori probability. In the time dimension, this method selects the measured value of the sensor to be detected at several previous moments as its time nearest neighbor. Then, the degree of abnormality is quantified according to the difference between the data of the sensor to be detected and the space-time neighbors, and the likelihood ratio test is used to judge whether the sensor to be tested is faulty. The results show that this method achieves a higher detection rate under the same false alarm rate than the existing one. For example, when the false alarm rate is 10%, this method improves the detection rate by 10% ~ 30%.