Indoor Localization Using Digital Trajectory Zhang Jingcong 5120309521

Indoor Localization Using Digital Trajectory Zhang Jingcong 5120309521 2015 -06 -02 1

1 OUTLINE 4 Background & Related work 2 My Research & Analysis 3 Experiment & Result Summary & Future work 2

Background & Related work TOPIC Indoor Localization Using Digital Trajectory 3

Background & Related work TOPIC Indoor Localization Using l Dead Reckoning te ura c c a o N Digital Trajectory 4

Background & Related work l GPS No accuracy TOPIC l Fingerprinting Indoor Localization - Online: construction of RM - Offline: localization based on RM Using Digital Trajectory AP 1 AP 3 Wi-Fi RSS(received signal No ac strength) cu AP(access point)s ra t AP 2 e 5

Background & Related work l GP No accuracy S TOPIC l Fingerprinting Indoor Localization Using Digital Trajectory channel noise reflection diffraction etc. accuracy density of fingerprints 6

My Research & Analysis target improve the accuracy with fewer fingerprint samples use Android smartphone embedded sensors to build a digital trajectory solution 7

My Research & Analysis WI-fi fingerprinting Particle Filter More accurate result Trajectory construction 8

My Research & Analysis trajectory building-using Android 9

My Research & Analysis trajectory building-using Android orientation & DETECTPV SENSOR S acceleration ALGORITHM: DEAD RECKONING INPUT: a location fix f=(fx, fy) acceleration a=(ax, ay, az) OUTPUT: trajectory tr={(x 1, y 1), (x 2, y 2), …} while new a do |a|=sqrt(ax^2+ay^2+az^2); if detect. PV(|a|)=true then calculate distance l; get theta θ; tr. add((xprev+l*sinθ, yprev+l*θ)); end digital trajectory using PV(peak-valley) detection to decide a step 10

My Research & Analysis Kalman Filter & Particle Filter weight State variance Particle Filter State Observation Trajectory Observation Estimation Kalman Filter N Wi. Fi fingerprint 11

Experiment & Result Trajectory building 12

Experiment & Result Kalman Filter truth observed filtered 13

Summary & Future work summary l. Generate a digital trajectory based on Android sensors l. Use Kalman fliter to get a better estimation Do experiment with real scene l Use Particle fliter instead of Kalman filter l Improve the Android program l future work 14

Reference l. Yang Liu, Marzieh Dashti, Jie Zhang. Indoor Localization on Mobile Phone Platforms Using Embedded Inertial Sensors l. Xiuming Zhang, Yunye Jin, Hwee-Xian Tan and Wee-Seng Soh. CIMLoc: A Crowdsourcing Indoor Digital Map Construction System for Localization 15

Q&A 16