A patient’s indoor positioning algorithm using Artificial Neural Network and SVM

초록

This paper proposes a patient’s Indoor Positioning Algorithm using Artificial Neural Network and Support Vector Machine (SVM). The proposed algorithm is ANN-SVM which combines Artificial Neural Network and Support Vector Machine to estimate the user position for IPS. The input data for the algorithm consists of Received Signal Strength Indicator and the location vector which is extracted by Access Point. The output is input weight and output weight. The input and output weight are processed by SVM with Room ID data. The last output is the estimated x and the room ID. According to the result of average class loss rate, SVM and ANN-SVM are 0.45 and 0.4, ANN-SVM has lower class loss rate by 0.05 than SVM. The accuracy rate of SVM and ANN-SVM are 65% and 70%. The ANN-SVM has more accuracy rate by 5% than SVM. © 2005 - Ongoing JATIT & LLS.

키워드