The principal objectives of this research are: to investigate the performance of different fingerprint-based WiFi Indoor Positioning Systems (IPS), analyse historical long-term data signals, detection of signal change points and outliers; then present an enhanced method that generates temporal based fingerprints.The proposed method consists of analysing signal strength profiles over time and detecting points at which the profile behaviour changes. This methodology can be used to dynamically adjust the fingerprint based on environmental factors, and with this select the relevant Wireless Access Points (WAPs) to be used for fingerprinting. The use of an Exponentially Weighted Moving Average (EWMA) Control Chart is investigated for this purpose. A long-term analysis of the WiFi scenery is presented and used as a test-bed for evaluation of state-of-the-art fingerprinting techniques. Data was collected and analysed over a period of 18 months, with over 840 different WAPs detected in over 77,000 observations covering 47 different locations of varying characteristics.A fully functional IPS has been developed and the design and implementation is described in this thesis. The system allows the scanning and recording of WiFi signals in order to define the generation of temporal fingerprints that can create radio-maps, which then allow indoor positioning to occur.This thesis presents the theory behind the concept and develops the technology to create a testable implementation. Experiments and their evaluation are also included.Based on the timestamp experiments the proposed system shows there is still room level accuracy, with a reduction in radio-map size.
|Date of Award||31 Dec 2015|
- The University of Manchester
|Supervisor||David Morris (Supervisor) & Martin Turner (Supervisor)|
- Indoor Positioning System