Wi-Fi Multi-Fingerprint Technique For High Accuracy Indoor Positioning
Determining position is very important because it is easier and faster to search for the best route to the destination with known position. Nowadays, more and more people move to urban areas and live in complex buildings. Indoor positioning, therefore, plays an important role for determining position for indoor areas and also in urban areas. Indoor positioning focuses on using smartphone to receive Wi-Fi signal due to its convenience and ease of operation. The Wi-Fi Fingerprint has been commonly used for indoor positioning. Nevertheless, Fingerprint technique has limitation in multipath rich indoor environment. Thus, this research focuses on improving the efficiency of Fingerprint technique by using multilayer of Fingerprint, named “Multi-Fingerprint Technique” in order to overcome the limitation of Fingerprint technique. Two types of Multi-Fingerprint technique were created in this research, which are (1) Patterned Fingerprint, radio map and converted radio map (or named distance map) with certain reference nodes were used for position determining; each reference node on distance map was calculated by the reversion of radio propagation model equation, and (2) Random Fingerprint, radio map and distance map with random reference nodes were used for position determining; each reference node was randomly defined by computer program. The four layers of radio maps and distance maps mentioned above were integrated (named Multi-Fingerprint map) and k-Nearest Neighbor (k-NN) algorithm was used for classifying position. Experimental results showed the difference in positioning efficiency among patterned distance map, random distance map, and Multi-Fingerprint map. Besides, root mean square error equation was used for measuring errors between real position and estimated position.
Keywords- Indoor positioning, Multi-Fingerprint technique, Received Signal Strength Indicator, Path loss model, k-Nearest Neighbor (k-NN)