Design And Implementation Of Mall Location Service System Based On Wi-Fi Technology

With development of our society and economy hits another milestone, the modern shopping environment is getting harder and harder to navigate. Due to its immobility and the low recognition efficiency, the experience of current indoor maps is extremely user-unfriendly. How to indicate accurate directions to shops or toilets and to provide fast navigation to closest emergency exits becomes an urgent issue needed to be solved for both customers and mall managers.Most shopping complexes are now widely deployed with wireless routers, and customers tend to use their smart phones to find locations, which mostly have Wi-Fi capabilities. Also, in 2016, Wi-Fi Alliance has announced the low-power, long-range "super Wi-Fi HaLow" technology, which is designated for IOT (Internet of Things) with working frequency at 900MHz band and strong capability passing through walls and obstacles. Through comparative analysis, we found Wi-Fi technology have significant advantages compared to traditional radio frequency, Bluetooth and magnetic on both positioning accuracy and feasibility in indoor environment. However, the traditional Wi-Fi fingerprint algorithm that has single model and requires to traverse and process all sample data cannot deliver acceptable performance in challenging environment such as shopping malls.To address the above problem, this thesis uses divide and conquer idea to deal with overall positioning problem with utilization of application scenarios and good generalization ability of support vector machine for small sample problems. In the offline training stage, we divide the whole issue into a number of models with appropriate number of samples, train support vector machine for multi-class models respectively, and save these trained models. In the online positioning stage, we predict location with corresponding model. By performance analysis, the method of this thesis compared to traditional ones shows great reduction in the amount of calculation, dramatic improvement in over-fitting prevention and major reduction in positioning error. In addition, this thesis introduces a design of the mall location services system of C/S model, which brings great convenience to users. Firstly, the overall architecture is introduced in System Server part. Followed by that is the elaboration of the implementation process and results of the system access control, positioning, data storage and system background management. In the Client part, Android devices are utilized as terminal. Firstly, the implementation of user management and position fingerprint scanning is discussed with several specific scenarios. SVG format is used to display indoor maps and construct the map information, Android is used to parse the SVG file and draw indoor maps. Compared to the traditional bitmap, SVG map reduces usage and prevents zoom distortion. The implementation and operation of other application functions is shown in the end.