Research On Calibration-Free Indoor Location Algorithm Based On Directional Signal Compensation

With the development of pervasive computing, location based services (LBS) are becoming m ore and more popular, particularly in emergency situations, such as in fire or the support facilities breaking down, we need an easy-to-setup and real-time locating system for timely location and rescue.Nowadays, there are many location method and system, such as the location of GPS, GSM, infrared rays, ultrasonic, Bluetooth and Wi-Fi. But those systems have some shortcomings for emergency scene. Some of them cannot use in indoor environment, some only can reach a short distance, some need to set infrastructure and some need huge amounts of preliminary work to acquire training data and it cannot meet the demand of time.To address this issue, we propose a location plan for emergency scene. We bring the directional antenna into indoor positioning and use directivity and penetration characteristic of it to enhance the location performance. We only need a few feature points to collect training data, avoiding large scale data collection work. We also do some research on the choice of the place of AP and number of training data collect points. At last, we propose a relative coordinates calibration method, avoiding manual calibration work at the later period.Then, we proposed a calibration-free indoor location method-DireMR based on directional signal compensation. Unlabeled data combined with a few labeled data as train data. We used signal compensation to avoid large scale data collection work. This method is similar to the general machine learning algorithms and it has train part and location part. The train part includes training data collection, the data feature extraction, data preprocessing and signal compensation. The location part includes location data collection, data preprocessing and the prediction of location coordinates.Finally, we design and develop an indoor location system.We carried out experiment in real environment, the result shows that it can reach high accuracy and can locate quickly and conveniently in the condition of less access points and label data.