Indoor positioning using Bluetooth

Abstract

As GNSS is not accurate enough for indoor use, several systems using alternative technologies for indoor positioning have been developed. Most prominent are the systems based on Wi-Fi. These rely on the strength of the Wi-Fi signals to determine the position of the user, either by using trilateration or by comparing to previously collected data. Using these techniques they provide accuracy enough for navigation, but access points are expensive and prone to changes in the environment. This thesis investigates whether using Bluetooth beacons with Bluetooth Low Energy is a viable alternative to Wi-Fi positioning.

This is done by studying the properties of Bluetooth signals, the effect of different environments and the accuracy achieved using several methods for estimating distance and position. Tests are performed to determine parameters of the logarithmic distance model and it's results are compared to a model which is independent of the environment. The method of DARL is compared to trilateration for determining a position from several distance estimations. Finally, using the earlier results, tests are done to determine the accuracy one can expect in a corridor using two different beacon configurations.

The strength of Bluetooth signals proved to be relatively stable up to a minute, but variations on a larger scale made it difficult to accurately determine the path loss exponent in the logarithmic distance model. This inaccuracy is probably what led to the two distance models performing approximately equal. DARL consistently provided more accurate position estimates than trilateration, most likely as a result of heavily favoring nearby beacons to those further away.

Using map matching in a corridor, average errors as low as 0.8 meters in the lengthwise direction were achieved using eight beacons to cover 16 meters of the corridor. The implication of this result is discussed with regards to an indoor positioning system. Further works should be aimed at reducing long term fluctuations in the received signal strength.