Scan statistics with local vote for target detection in distributed system

Abstract

Target detection has occupied a pivotal position in distributed system. Scan statistics, as one of the most efficient detection methods, has been applied to a variety of anomaly detection problems and significantly improves the probability of detection. However, scan statistics cannot achieve the expected performance when the noise intensity is strong, or the signal emitted by the target is weak. The local vote algorithm can also achieve higher target detection rate. After the local vote, the counting rule is always adopted for decision fusion. The counting rule does not use the information about the contiguity of sensors but takes all sensors' data into consideration, which makes the result undesirable. In this paper, we propose a scan statistics with local vote (SSLV) method. This method combines scan statistics with local vote decision. Before scan statistics, each sensor executes local vote decision according to the data of its neighbors and its own. By combining the advantages of both, our method can obtain higher detection rate in low signal-to-noise ratio environment than the scan statistics. After the local vote decision, the distribution of sensors which have detected the target becomes more intensive. To make full use of local vote decision, we introduce a variable-step-parameter for the SSLV. It significantly shortens the scan period especially when the target is absent. Analysis and simulations are presented to demonstrate the performance of our method.

A Clustering Algorithm to Improve the Scan Statistic in Sensor Detection Systems

“…where (16) follows because in this example all clusters have the same β(C, t|l e,j ). Since clusters are nonintersecting in this example, which means that {T (Z C ) : C ∈ C} are independent given the event {L e = l e,1 }, we obtain…”

“…It is interesting to observe that |C poi | = 4 and |C ∼> | = 5 and the removal of just one cluster ({2}) makes C poi perform better than C ∼> . The better performance of C poi can also (10,40), (20,20), (11,16), (40,40), (40,50), (50,40), and (50,50) are illustrated with dashed lines. These and other clusters not illustrated form the final cluster set C poi .…”

“…Although suboptimal, the scan statistic has been successfully used to detect anomalies in georeferenced data [9], [13]- [15] and many authors have recognized its value in distributed sensor detection systems [4], [12], [16]- [20]. Here, because of the signal attenuation between emitter and sensors, the change in the distribution of the measurements collected by sensors near the emitter can be considered an anomaly in the set of all measurements, motivating the use of the scan statistic.…”