Crowdsourcing-based Indoor Map Construction And Localization Algorithm

In recent years,with the widespread popularity of smart mobile devices and WLAN,the demands for accurate location services become more and more intense.The outdoor location technology which is represented by GPS has been unable to meet the demand for accurate and convenient indoor location services.However,most of the existed indoor positioning technologies dependent on specific hardware conditions,so it is difficult to promote them in practice.Indoor positioning technology based on WLAN(Wireless Local Area Network)and RSS(Received Signal Strength)become a hot topic in the field of indoor positioning.Because these methods are entirely based on the existing hardware facilities while can still meet the localization need for most indoor location services.Apparently,WLAN-based indoor positioning technology needs both floor plan and indoor signal map.Traditionally,to construct an indoor floor we must do the ”site survey” work which is labor-intensive.On the other hand,constructing the signal map is also very complex due to the instability of the indoor signal propagation environment.The propagation-model based signal map cannot meet the needs of indoor positioning accuracy and the fingerprint-based indoor positioning methods may suffer the problem of high labor cost on fingerprint construction phase.Based on this background,in our paper,we first analysis both the advantages and disadvantages of existing work,and propose the solution of indoor map construction system by combining crowdsourcing and indoor dead reckoning.On the research of crowdsourcing,we first point out the challenges and problems of the existed work,and propose crowdsourcing based map construction algorithm.To solve the low accuracy problem of fingerprint database caused by the sparse of spatial distribution of Wi-Fi samples,we raise our model based Wi-Fi fingerprint construction algorithm.In addition,we also redesign the step counter which is based on walk frequency and the experiment result shows high accuracy.Finally,we implement the map construction and localization system,and evaluate it with the related works.In our work,we construct the indoor digital map by combining crowdsourcing with mobile computing technology,and the map can also be used in indoor localization.Through our work,the indoor map and indoor location fingerprint database can be constructed with low labor cost.In order to verify the accuracy of our system,we evaluate our prototype as well as relative algorithms by a system level experiment,and the result shows the effectiveness.