Hybrid Nearest Neighbour and Feed-forward Neural Networks Algorithm for Indoor Positioning System

Accurate localization of people and mobile devices in building plays important roles in order to deliver helpful location-based services such as direction and object finding, and content delivery. This paper presents a hybrid feed-forward neural networks (FNNs)-based indoor localization system trained by means of recently developed metaheuristic algorithm. The received signal strength (RSS) from the access points (APs) is employed as input to the proposed algorithm to estimate the user location in the wireless local area network (LAN) environment. The performance of developed hybrid
$\pmb{k}$
NN-FNNs algorithms were compared with the basic
$\pmb{k}$
-NN algorithm in term of error distance achieved. The comparative study has shown that the developed hybrid
$\pmb{k}$
-NN-FNNs algorithm is more efficient when estimating the user location with accuracy of 86.39 percent in comparison to its predecessor algorithm (with 69.67 percent) as the error distance was minimized using metaheuristic algorithm.