Wireless Location in Indoor Sensor Networks

The thesis aims at the development of a realistic simulation environment for an indoor wireless personal area network (PAN) positioning system, based on the received signal strength (RSS). Positioning system based on RSS will be a popular choice in the wireless internet of things (IoT), due to its simplicity, no need for a special hardware and therefore, possibility of working with different protocols and technologies.

From pathloss characterisation of the area, a practical approach has been developed to the simulation of statistical variations of the received signal strength in wireless indoor PAN using Gamma and log-normal distributions. The characterisation of the indoor location environment (either by measurement or by simulation) requires determination of only a basic pathloss model (a relationship between the average received power, Rx, and the distance) and the average standard deviation, ?, of the received power. Typical data for various indoor environments can also be used. This is a far simpler approach than that used by many other researchers, who described pathloss statistics by separate terms due to large-scale fading, small-scale fading, and antenna rotation effects. The thesis demonstrates that for a typical office indoor environment, a single statistical pathloss model is equally accurate in location application as the more complex models. A novel mathematical procedure has been developed to find the parameters of the Gamma distribution from the basic pathloss model, which yields the required statistical variation of the received signal strength. The same procedure can be applied to calculate parameters of Nakagami distribution, which is directly related to Gamma by square root operation. A comparison of the pathloss prediction by the log-normal and Gamma distributions with practical measurements is presented. Location simulator has been developed which essentially consists of two main elements: generation of randomly varying received power and estimation of the device location from the RSS values of the beacons. The simulator was used to study various versions of the location algorithm using statistically varying RSS information from multiple beacons for two different indoor environments. Also, the effect of the number of beacons on the average of the location error was studied.

A practical Zigbee location system has been developed and deployed in indoor environments under study. Results from the location simulator were compared with practical measurements, proving the validity of the simulation approach. The results have shown that Gamma distribution is slightly better than the log-normal distribution in predicting positioning errors.