Improved localization using Kalman filter on estimated positions

In this paper we present a computational-efficient two-phases model for localization and tracking based on Kalman filter. A first estimate of target position is obtained via Super MDS algorithm only using noisy distance measurements, then location information is refined via a classic Kalman Filter exploiting the noisy acceleration of the target. The main scientific contribution of this paper is to show that, although the information theory proves that such a sequential approach is sub-optimal, the performance is accurate enough even with high-noisy acceleration measurements. This fact suggests that in the vast majority of use cases is possible, taking advantage of its mathematical simplicity, to employee this two-phase model neglecting its sub-optimality.