A Community-Based Cooperative Anomaly Detection System by the Synergy of Mobile Sensing and Delay Tolerant Networks

This paper proposes a novel cooperative anomaly detection system that uses mobile sensing and delay tolerant network (DTN) technologies. As an example application, the anomaly detection system in which vehicles detect dangerous states of the road that could cause traffic accidents and/or traffic jams is considered. Assuming that the predicted road states by the anomaly detection include some prediction errors, a hidden Markov model (HMM) is introduced for representing the spatial relationship among the road states. The EM algorithm is employed to estimate the HMM parameters such that the Viterbi algorithm is applied to a target predicted road sequence. In addition, the reliable road states shared with other vehicles by the DTN technology are combined with the locally predicted sequence to perform trellis pruning in the Viterbi algorithm as the cooperative data processing, which further improves the accuracy of road state estimation.