Multi-mode Data Fusion For High Accurate Indoor Positioning Methods And System

In recent years, with the rapid development of mobile Internet and intelligent terminals, location based service(LBS) are widely used in people's daily life. Nowadays, outdoor GPS based location is very mature, and indoor location technologies get lots of attention from academics and industry. Currently, there are a lot of indoor location technologies, many technical solutions have been used in different scenarios of indoor positioning applications, such as motion sensors, Wi-Fi, Bluetooth, magnetism, LED, etc. But there isn't a single location technology can meet the requirements of both high accuracy and low cost. Wi-Fi based location methods can provide location service with low cost by using the existing environment AP. But due to the differences in terminal devices, the location accuracy has a big gap in different devices. The long time location period cannot provide smooth location trajectory. On the other hand, motion sensors based location methods are also popular in lots of applications. But the error accumulation limits the long term location service. Therefore, in this article, we firstly make use of fingerprint features fusion to eliminate the differences of Wi-Fi fingerprint in devices, and then we combine the Wi-Fi based location with the motion sensor based location method to improve location accuracy and generate more smooth trajectory.The main contents of this work include the following three parts:1. We propose a Wi-Fi fingerprint features fused indoor localization method. For the diversity of device degrades the performance of location estimation, our method utilizes the processing of removing linear effect to extract fingerprint features, combining that applies AP effective factor to computer fingerprint distance, then implements to eliminate the effect about the device diversity decreasing the localization accuracy. Experimental results show that, this method can adaptively meet the diverse property of different devices, and obtain robust and accurate location estimation effect.2. We propose a multi-mode sensors and Wi-Fi fingerprint fused localization method. For fingerprint localization waving and motion sensor accumulated error leads the drift problem of indoor location and movement trace, our method employs Wi-Fi fingerprint achieving the starting point and correcting move locations, fusing that utilizes multimode motion sensors to get the velocity and direction between adjacent locations, then implements to predict the continuous move locations and trajectories. Experimental results show that, this method can timely correct the drift phenomenon of location and trace, and obtain precise user's current location and movement trace.3. Basing on our proposed fusion location method, we design and implement an indoor location estimation system which fusion Wi-Fi fingerprint location and motion sensor location, including a fingerprint calibration client, a Wi-Fi localization server and a fusion localization client. Meanwhile, we packed our location algorithm to be a standard interface SDK, which can be easily provided for third-party companies to develop indoor LBS applications, and validate the practicality and usability of our SDK in lots of shopping malls.