Indoor positioning has been around for quite a while. Primarily being handles by Google maps and few other startups.
 In this post we will discuss basics of Indoor Positioning. First we have to mark the indoor location to be mapped, we have mark the region for radio signals. This is termed as Fingerprinting. 
Phase 1 : Training Phase 
Make Radio map. We place 3 detectors (minimum) and take note of signal strength from three detectors.
 We divide whole floor in “n x n” squares and mark each square with radio signal strength.

Detectors : D1; D2; D3
Signal Strength = RSSI1 ; RSSI2 ; RSSI3
Squares = S1 to Sn

Fingerprinting Table

Once we have this table, we freeze readings of this table. Now when object in question moves around, we compare its signal strength with that of this table and estimate its position.

Phase 2 : Position Estimation Phase
 The position in question will be the one that is associated with the fingerprint of the best match or the geometric median of the positions of the K-closest n fingerprints.
 When calculating the position, two approaches are used, static Bayesian estimation and estimation with a motion model. 
 1) Bayesian Estimation
 In static Bayesian estimation, the position is estimated based only on the current RSSI measurements that are collected from beacon nodes in range. 
 2) Motion Model Estimation
 When the motion model is utilized, earlier positions and the velocity are applied in the calculations. When initializing the estimation a uniformly distributed prior such is used since no previous data is available in the firs iteration.