From RSSI to CSI文献No Need to War-Drive Unsupervised Indoor Localization.pdfVIP

No Need to War-Drive: Unsupervised Indoor Localization He Wang Souvik Sen Ahmed Elgohary Duke University Duke University EJUST, Egypt Moustafa Farid Moustafa Youssef Romit Roy Choudhury EJUST, Egypt EJUST, Egypt Duke University ABSTRACT high location accuracy because the contexts vary at finer spa- We propose UnLoc, an unsupervised indoor localization scheme tial granularity. For instance, 5m location error outdoors may that bypasses the need for war-driving. Our key observation is still indicate the same street, but 5m indoors may mean two that certain locations in an indoor environment present iden- different aisles in a grocery store. An inventory-management tifiable signatures on one or more sensing dimensions. An application may very well require aisle-level precision. (2) While elevator, for instance, imposes a distinct pattern on a smart- such precision is attainable with pervasive WiFi systems, they phone’s accelerometer; a corridor-corner may overhear a unique come at a prohibitively high cost, mostly in the form of metic- set of WiFi access points; a specific spot may experience an ulous (signal) calibration. Such calibration is not necessarily a unusual magnetic fluctuation. We hypothesize that these kind one-time cost since RF fingerprints could change, perhaps due of signatures naturally exist in the environment, and can be to changes in layout and objec