| Recently,the development of mobile network and intelligent terminals enables a large collection of applications that facilitate people's daily life.Indoor navigation based on mobile phones is one of the emerging applications.For example,intellectual guide in shopping malls or supermarkets;looking for a meeting room in an unfamiliar office building;searching cars in a large parking lot,etc.Global Positioning System is well known as it is widely used in outdoor navigation system.However,GPS is unable to realize the reliable locating in complex indoor environment because of the signal attenuation and multipath effect.Therefore,designing an indoor navigation system which is reliable and with high real-time to meet the growing demand for indoor positioning service has great significance.Existing indoor navigation systems(such as indoor navigation systems based on Wi-Fi and Bluetooth)mostly rely on network infrastructure(such as wireless AP and Bluetooth base stations,etc.)and large geomagnetic fingerprints databases.However,these systems can do nothing without the network infrastructure.In response to these issues,this article puts forward a novel indoor navigation system based on geomagnetic fingerprints which can easily realize indoor navigation by using only a smartphone.Compared with the indoor navigation system based on Wi-Fi and Bluetooth,our system is freed from network infrastructure and it can be widely applied in different scenarios because of the flexibility and low implementation cost.In order to realize the navigation system mentioned above,this paper solved the following several challenging problems:(1)Fingerprints database is also the precondition for our system.However,the construction of the fingerprints database in some large buildings will be very complex.This paper applies crowdsourcing to the construction of the fingerprints database which can greatly reduce the workload of database's construction.(2)In order to realize the localization,this paper designs a matching algorithm for geomagnetic time series according the result of observational study in geomagnetic time series.The algorithm calculates the similarity between the geomagnetic time series collected from current path over a period of time and that in the database.Then,the time series which has the highest similarity is mapped to map coordinates so as to realize localization.(3)Strong magnetic anomaly points more or less exist in buildings which visually display as peak or trough in geomagnetic time series images.Therefore,it is easy to identify these points.This paper regards these points as ‘feature points' which are used to calibrate the positioning results.The calibration algorithm improve the precision of the navigation system.The characteristic of navigation system presented in this paper is that it only uses the geomagnetic signals in buildings to implement localization and navigation and is free from network infrastructure.Besides,the system makes use of crowdsourcing to solve the heavy work of map construction.This paper finally implements the navigation system on an Android smartphone with a magnetic sensor.And we invite four volunteers to take part in the performance evaluation of the system in a school building of 700 square meters.The experiment results show that at 95 percent of time the space error is below 4m,and the maximum error is less than 10 m and the average space error is below 3m.This proves that the system can provide real-time positioning and navigation for users.And it performs particularly well in localization accuracy.In addition,this paper studies the characteristics of magnetic time series and puts forward an algorithm based on dynamic time warping to measure the similarity between two magnetic time series with different length,it can broaden thoughts for the research of indoor navigation system based on geomagnetic fingerprints. |
PDF Full Text Request