Modeling HMM Map Matching Using Multi-label Classification

Map matching deals with GPS coordinates to corresponding points or segments on a road network map. The work has various applications in both vehicle navigating and tracking domains. Traditional rule-based approach for solving the problem yielded great results. However, its performance depends underlying algorithm Mathematical/Statistical models employed approach. For example, HMM Matching O(N2) time complexity, where N is number of states Hidden Markov Model. techniques large order complexity are impractical providing services, especially within time-sensitive applications. This due their slow responsiveness critical amount computing power required obtain paper proposed novel data-driven projecting trajectory onto network. We constructed supervised-learning classifier using Multi-Label Classification (MLC) technique Analytically, our yields O(N) suggesting that better running when applied matching-based which response major concern. In addition, experimental results indicated we could achieve Jaccard Similarity index 0.30 Overlap Coefficient 0.70.