RETRACTED ARTICLE: Positioning of WiFi devices for indoor floor planning using principal featured Kohonen deep structure

Abstract

With the evolution of Internet and mobile intelligent devices, position based servers has become the borderline in the research area of information technology. WiFi technology is extensively exploited for efficient indoor floor planning due to low erection cost, suitable access, ease of expansion and popularization, etc. For effective indoor floor planning, Indoor Positioning System has to be designed initially so that objects are said to be located inside building wirelessly with minimum computational time and overhead. To explore this design, Ipin2016 Dataset is used, Principal Feature Enhanced Sampling with Auto Encoder for feature selection and built positioning model based on Kohonen Deep Structure. The method is called as Positioning of Wi-Fi devices for indoor floor planning using Principal Featured Kohonen Deep Structure (PF-KDS). First, spatial data analysis is carried out using Principal Feature Enhanced Auto Encoder algorithm. With this algorithm, principal features are first extracted, so that dimensionality reduction is said to achieve, therefore reducing the complexity involved in positioning. Next, with the dimensionality reduced features, Kohonen Self Organizing Deep Structured Learning algorithm is designed. In this algorithm, list of single medium access control (MAC) address and analogous Receiving Signal Strength (RSS) is recorded by considering a new path loss model including walls’ influence on RSSI, therefore improving the positioning accuracy. Our results indicate that combination of Principal Feature Enhanced Sampling with Kohonen Deep Structure provides high positioning accuracy with minimum time and overhead for Indoor Positioning.