Tack: Learning Towards Contextual and Ephemeral Indoor Localization - - PowerPoint PPT Presentation
Tack: Learning Towards Contextual and Ephemeral Indoor Localization With Crowdsourcing Liyao Xiang ECE Dept. Nov. 24, 2017 Indoor Localization Traditional localization infrastructure is costly. Most user devices are common smartphones.
2
3
4
5 Zi,t
Dij,t
Zj,t Zk,t
6 Zi,t
Dij,t
Zj,t Zk,t
7 Zi,t-2
Dij,t-2
Zj,t-2
time window = 3
Zk,t-2
Dik,t-2
Zi,t-1
Dij,t-1
Zj,t-1 Zk,t-1
Dik,t-1
Zi,t
Dij,t
Zj,t Zk,t
Dk,t-2 Dk,t-1 Di,t-2 Dj,t-2 Di,t-1 Dj,t-1
8 Zi,t-2
Dij,t-2
Zj,t-2
time window = 3 time window = 3
Zk,t-2
Dik,t-2
Zi,t-1
Dij,t-1
Zj,t-1 Zk,t-1
Dik,t-1
Zi,t
Dij,t
Zj,t Zk,t
Dk,t-2 Dk,t-1 Di,t-2 Dj,t-2 Di,t-1 Dj,t-1 backward propagation forward propagation forward propagation
9
Accelerometer Magnetometer Bluetooth transmitters Step Counter Dead Reckoning Local Estimator Position Estimates Encountering User Other Users Inference algorithm
10
11
Experiment Setting: 7 beacons, 7 users
HMM Window = 3 Window = 5 1 2 3 4 5 Mean error (m) Mean Error for All Users in Different Settings. 5 Users 7 Users
12
13