wifi based indoor localization system by using weighted path
TRANSCRIPT
WiFi Based Indoor Localization System by Using WeightedPath Loss and Extreme Learning Machine
Han ZouNanyang Technological
UniversitySingapore, 639798
Hao JiangNanyang Technological
UniversitySingapore, 639798
Lihua XieNanyang Technological
UniversitySingapore, 639798
1. WIFI BASED INDOOR LOCALIZATIONSYSTEM
The methodology of our WiFi based indoor localiza-tion system is built upon passive cooperation of occu-pants only which does not interrupt the daily lives ofthem. Instead of modifying the hardware or software ofoccupants’ mobile devices, we upgrade the software ofthe existing commercial WiFi access points (APs) in theindoor environment to WiFi sniffers, which can detectthe received signal strength (RSS) of each mobile device.The RSS and MAC addresses of mobile devices are thensent to a location processor. The location processor willleverage on appropriate localization algorithms to figureout the position of each mobile device and thus its us-er. Based on our experimental results, our system canprovide around 2m localization accuracy consistently indifferent indoor scenarios.
The deployment components of our system are onlynumbers of software upgraded WiFi APs which can cov-er the entire competition testbed, and a laptop as thelocation processor.
2. LOCALIZATION ALGORITHMSWe proposed the Weighted Path Loss (WPL) approach
in [1], which can be classified as a centralized model-based localization algorithm. WPL works as follows:The distance between the mobile device and each WiFiAP is calculated based on a modified indoor path lossmodel in the first place. Then the estimated location ofeach mobile device is obtained as the summation of eachWiFi AP’s weighting factor (reciprocal of the distancebetween the mobile device and each WiFi AP) multi-
Permission to make digital or hard copies of all or part of this work forpersonal or classroom use is granted without fee provided that copies arenot made or distributed for profit or commercial advantage and that copiesbear this notice and the full citation on the first page. To copy otherwise, torepublish, to post on servers or to redistribute to lists, requires prior specificpermission and/or a fee.Copyright 200X ACM X-XXXXX-XX-X/XX/XX ...$5.00.
plied by its physical location, provided all the physicallocations of the WiFi APs are known. As shown in [1],WPL approach can provide higher localization accura-cy, faster estimation and more robustness than othermodel-based approaches. It is suitable to deliver indoorpositioning services with high localization accuracy inlarge open indoor environment.
Besides the model-based approach, another catego-ry of localization algorithm is fingerprinting-based ap-proach. We also proposed a fingerprinting-based local-ization algorithm which is based on Extreme LearningMachine (ELM) in [1]. ELM is a kind of machine learn-ing algorithm based on a Single-hidden Layer Feedfor-ward neural Network architecture. It has been proved toprovide good generalization performance at an extreme-ly fast learning speed. It consists of two phases: offlinecalibration phase and online localization phase. Duringthe offline calibration phase, WiFi RSS fingerprints arecollected at several known locations in the first place.Then, these WiFi RSS fingerprints and their physicallocations are adopted as training inputs and trainingtargets respectively to build up an ELM model for onlinelocalization. During the online localization phase, afterfeeding the current measured WiFi RSS data of eachmobile device into the ELM model, the output given byELM is the estimated location of each mobile device. Asshown in [1], ELM has tremendous advantages in offlinetraining time and online localization accuracy. More-over, it can provide outstanding localization accuracyin complex indoor environment.
Appropriate localization algorithms will be integratedand leveraged to adapt various indoor scenarios duringthe competition.
3. REFERENCES[1] H. Zou, H. Wang, L. Xie, and Q.-S. Jia, ”An RFID in-
door positioning system by using weighted path loss andextreme learning machine,” in Cyber-Physical Systems,Networks, and Applications (CPSNA), 2013 IEEE In-ternational Conference on, 2013.