Sensor-Assisted Wi-Fi Indoor Location System for Adapting to Environmental Dynamics Yi-Chao Chen, Ji-Rung Chiang, Hao-hua Chu, Polly Huang, and Arvin Wen Tsui Intelligent Space Laboratory National Taiwan University Presenter: Willy, Ji-Rung Chiang
10/10/2005MSWiM Outline Background Problems Approach –System Architecture Experiment Results Related Works Conclusion
10/10/2005MSWiM Background WiFi-based indoor location system –Use existing IEEE network infrastructure –Meter-level error Fingerprinting method –Offline training phase: Record RSSI (Received Signal Strength Indicator) from different APs at some sample points & build radio map –Online localization phase: Matching sampled points on the radio map with the closest RSSI values to the target
10/10/2005MSWiM Problems Calibration effort –2 man-hours for a 1000m 2 environment Instability of RSSI –Environmental dynamics reduce positioning accuracy Humidity level People presence and blocking Open/Close Door
10/10/2005MSWiM Instability of RSSI RSSI on Different Humidity RSSI on People Blocking
10/10/2005MSWiM Approach Sensor-assisted Online Calibration –Reduce Calibration Effort RFID: label RSSI samples with location –Adapt to Dynamic Factors Environmental sensors: label samples with environmental condition –Humidity, people presence, open/close doors
10/10/2005MSWiM Sensor-Assisted Online Calibration x i =x 0 + (t i –t 0 ) * v x y i =y 0 + (t i –t 0 ) * v y v x =(x 4 –x 0 ) / (t 4 –t 0 ) v y =(y 4 –y 0 ) / (t 4 –t 0 ) where i = 1~3 (x 0, y 0 )(x 4, y 4 ) SS 3 t3t3 SS 2 t2t2 SS 1 t1t1 t0t0 t4t4 (x 1, y 1 )(x 2, y 2 )(x 3, y 3 )
10/10/2005MSWiM System Architecture Radio map Context-aware Radio maps Location Calibration Online RSSI Sample Filter Online Training Engine Adaptive Location Estimation Engine Environment sensors (e.g., humidity sensor) RFID-assisted location estimation Labeled Online RSSI Samples Client RSSI values Location Estimation Sensors Query current state of environmental condition Select a radio map Sensor-assisted Sample Collection Phase Online Calibration Phase Adaptive Localization Phase Environment condition
10/10/2005MSWiM CORRIDOR One Example Location Online RSSI Sample Filter Environment sensors (e.g., humidity sensor) RFID-assisted location estimation Labeled Online RSSI Samples Client RSSI values Sensor-assisted Sample Collection Phase Environment condition Sensors Radio map Context-aware Radio maps Online Training Engine Labeled Online RSSI Samples Online Calibration Phase Adaptive Location Estimation Engine Environment sensors (e.g., humidity sensor) Client RSSI values Location Estimation Query current state of environment condition Select a radio map Adaptive Localization Phase Radio map Context-aware Radio maps
10/10/2005MSWiM Experimental Results
10/10/2005MSWiM Number of Trace vs. Accuracy
10/10/2005MSWiM Impact of Open/Close Door
10/10/2005MSWiM Impact of People Blocking Corridor
10/10/2005MSWiM Impact of Humidity
10/10/2005MSWiM Related Works Reducing Calibration Effort –X.Chai, Q.Yang, “Reducing Calibration Effort for Estimation Using Unlabeled Samples”, Percomp 2005 Reduce the amount of sample points of radio map by interpolation, but it is still in need of offline manual calibration progress. Effects of Environmental Dynamics –J. Yie, Q. Yang, L. Ni, “Adaptive Temporal Radio Maps for Indoor Location Estimation”, Percomp 2005 Under the assumption of the changes in the dynamic factors follow some predictable temporal patterns, they placed emitters and sniffers to learn the temporal relationship. However, not all factors are temporally predictable, e.g. people presence and blocking.
10/10/2005MSWiM Conclusion “Calibration Effort” and “Effects of Environmental Dynamic Factors” are two majors problems of modern WiFi location systems. In this paper, we proposed a sensor-assisted method to solve both of them. The current method is restricted to be deployed in corridors rather than within rooms. And the people blocking case is still limited.
10/10/2005MSWiM Thank You! please contact me at Willy Chiang