Online Target Localization Using Adaptive Belief Propagation in the HMM Framework

This letter proposes a novel adaptive sample space-based Viterbi algorithm for target localization in an online manner. The method relies on discretizing the target's motion space into cells representing finite number of hidden states. Then, most probable trajectory tracked is computed via dynamic programming Hidden Markov Model (HMM) framework. proposed uses Bayesian estimation framework which neither limited to Gaussian noise models nor requires linearized model or sensor measurement models. However, HMM-based approach can suffer from poor computational complexity scenarios where states increases due high-resolution modeling large space. To improve this complexity, belief propagation with low sequentially, reducing required resources significantly. inspired by $k$ -d Tree (e.g., quadtree) commonly used computer vision field. Experimental tests using ultra-wideband (UWB) network demonstrate our results.