The applicability of MCMC inference to track multiple targets within unthresholded measurements

This paper presents a Markov Chain Monte Carlo (MCMC) method for multitarget tracking within raw measurements. We derive the optimal proposal density so that the raw and unthresholded measurements could be used to generate approximating particles appropriately. However, because the optimal proposal density has exponential complexity, we apply the Gibbs sampler, the well-known MCMC method, to sample from the optimal proposal density and relieve sampling burden. Simulation results show that our strategy for using the Gibbs sampler could reach to a good compromise between accuracy and computation expense.