Support Vector Machine Algorithm-based Wireless Indoor Positioning System

With the rapid development of mobile applications,positioning information plays an increasingly important role in people's lives.In particular,providing accurate location information in indoor environments will further improve the quality of life and improve the convenience of living.Global Positioning System(GPS)and BeiDou Navigation Satellite System(BDS)have been able to meet the needs of outdoor positioning and navigation,but indoor positioning technology has not been able to meet the needs of the general public.At present,the coverage of wireless networks in public places in cities continues to increase,and algorithms based on Received Signal Strength Indication(RSSI)information have ample room for development.Wi-Fi-based indoor positioning technology can achieve costs low,high scalability,high flexibility indoor positioning system.In view of the technical difficulties in the Wi-Fi-based indoor positioning system,this thesis uses a machine learning method based on support vector machines to construct indoor positioning methods and technical solutions.The indoor positioning program mainly includes offline training phase and online prediction phase.The main work of the offline phase is to collect reference point fingerprint information and train the prediction model.The main work in the online phase is to collect fingerprint data in real time and predict the real-time position according to the prediction model.Because the indoor environment is complex and prone to change,there may be deviations between the training sample data collected by the mobile device during the off-line phase and the data collected during the on-line phase,resulting in inaccurate prediction positions.To solve this problem,this thesis proposes a training program for training data and real-time data acquisition: In the off-line phase,the received training data is filtered through the designed algorithm to remove the noise data to improve the quality of the training model of the support vector machine.In the online phase,the RSSI information is collected in real time through multiple consecutive receptions,and the impact prediction is removed through the designed screening strategy.Resulting Access Point(AP)to Improve Support Vector Machine Prediction Accuracy.Based on the UJIIndoorLoc dataset published by the University of California Irvine(UCI),the experimental results show that the wireless indoor positioning method based on Support Vector Machine(SVM)proposed in this thesis has higher positioning accuracy.And it has a good universality.