Research On Mobile Localization Algorithm Based On Smart Phones

Increasing demand on temporal-spacial location information arises with theintegration of human activities. Take tracking and localization in military for example.Besides, localization and navigation can be used on emergency rescue, traffic controland human localization in public place. It is quite obvious that localization andnavigation technologies are of great importance. We try to propose an end-to-endalgorithm with higher localization accuracy, lower dependency, wider coverage,stronger environmental adaptation, higher robust design in this thesis. What is more,energy consumption is also taken into our consideration.In this thesis, I try to navigate a person to the desired person in public place usingsmart phones. The application background is described as follows. It is reallyfrustrating to find a person in public place sometimes, such as library, hotel, orshopping mall. Difficulty arises when we do not know the exact position at the time.Even If we know, finding a person in unfamiliar place may be troublesome. Perhapsmaps and building guidance can help us to find the circumjacent position, but we cannot find guidance whenever and wherever.80%of users in our optional poll think theideal way to find a person is to have an end-to-end navigation service to accompanyand navigate us there.I design the localization and navigation algorithm based on CrowdSourcing usingsmart phones in this thesis. In the maintenance phase, client phones periodicallyacknowledge walking trails of different individuals, as well as how they encounter inspace-time. After uploading information to the cloud server, a global routing graph canbe computed with the help of cloud server. Therefore, the most efficient route betweenany pair of individuals can be obtained. In the operation phase, whenever user A asksthe cloud server for routing information, the server will send the steps sequence to userA’s smart phone. With the help of these sequences, user A could get his desired person.Visual recognition algorithm may help when they can not identify each other. An arrowwill display us with the desired person. This algorithm is also suitable for indoorlocalization or navigation in three-dimensional environment. All we need in the algorithm is a video beacon with fixed position. We do not rely on GPS, WiFi orWar-driving, leading to less dependency on external elements. All users need to do is tofollow the arrow on his phone. Experiments show the algorithm has an acceptableaccuracy of8m.