Real-Time Trajectory Similarity Processing Using Longest Common Subsequence
Jiafeng Ding, Junhua Fang, Zonglei Zhang, Pengpeng Zhao, Jiajie Xu, Lei Zhao · 2019
Driven by the fast development of mobile internet and mobile devices, huge volumes of trajectory data describing the spatial-temporal information of moving objects are currently generated. Trajectory similarity processing is playing a critical role to explore valuable information, such as the patterns of human activities and behaviors. However, most of the existing works optimize this operation in an off-line manner, which is unable to provide timely feedback for time-sensitive applications. In this paper, we present a new real-time processing framework for trajectory similarity measurement, with practical algorithms to reduce the computing resource cost and guarantee the correctness of the result. Our proposal includes a cost-effective framework to ensure the correctness of trajectory similarity processing, a stream partition algorithm to maximize the processing throughput, a lightweight Top-k computing mode to cut off unnecessary network cost. Empirical studies on real-world stream applications validate the usefulness of our proposals and prove the huge advantage of our approaches over state-of-the-art solutions in the literature.