Asynchronous periodic patterns discovery for moving objects

Periodicity is very common for moving objects, and current periodicity detection often mining the synchronous periodic patterns for moving objects. However, the activity of an object can be affected by outside influence greatly and synchronous periodicity detection can not find patterns with misaligned presence. Therefore, the results are often imprecise due to the random noise intervened into the activity sequence. In this paper, an approach of asynchronous periodic patterns discovery (APPD) for moving objects is proposed to solve the shifted and distorted in the periodic sequence. Firstly, a density based algorithm is given to find object's activity. Secondly, a flexible pattern structure is given to present the periodic patterns whose occurrence may be shifted due to disturbance. Thirdly a single period pattern detection algorithm is proposed to find asynchronous periodic patterns. In addition, we study the two interesting variants in the algorithm, and demonstrate how the algorithm can be adapted for the variants. Experiments on real data set show the efficiency, effectiveness of the proposed algorithms.