This report studies mobile indoor localization using Kalman filters to estimate the current state in a hidden Markov chain. The optimal Kalman filters has been a very successful tool in a wide variety of applications and has also proved useful in localization. The report introduces and then derives the Kalman filter in the context of the larger class of Bayesian filters.

Location data from a faux office is simulated from a variation of the random way-point mobility model using the Heat equation for path finding.

An heuristic attempt to improve the Kalman filter by reusing old inferred positions is done with ambiguous results. The Kalman filter can be slightly improved although it is not worth the extra computational effort.