Skip to content

An illustrative example demonstrating how to apply the HsMM Forward Recursion algorithm for indoor user location region positioning.

Notifications You must be signed in to change notification settings

shuai000/HsMM-ForwardRecursion

Repository files navigation

An HsMM-based forward recursion algorithm for real time indoor localization

(Development Python Code)

Author

Description

This code is developed as an illustrative example for the simulation of applying the HsMM-FR algorithm for indoor user region localization

Dataset specifications

The dataset consists a training and testing folder with received signal strength (RSS) data, obtained from a ray-tracing software (Wireless Insite)

There are K=8 anchor nodes and N=12 regions, see FigureSimu Note that this is slightly different from the figure used in the paper

  • In the folder _training, it contains 8 * 12 = 96 separate .p2m files, each file corresponds to a RSS data collected from a grid number of user locations (coordinate is given in the .p2m file) with respect to a certain Anchor node

  • In the folder _testing, it contains 8 separate .p2m files corresponds to a RSS data set collected from a moving user trajectory locations (coordinate is given in the .p2m file) with respect to each Anchor node

  • The ground truth location region of the user is shown in Figure 4, with st_ground = [[1, 8], [2, 18], [3, 23], [4, 36], [5, 46], [6, 62], [8, 78], [7, 120], [9, 129], [10, 138]]. To interpret it, it means the user stays in region 1 for 8 time steps, then transits to region 2 with a sojourn time of (18-8)=10 steps, then transits to region 3 and stay there for (23-18)=5 steps, ...

Naming Convention (training dataset)

  • Notation in the .p2m file: **power.t001_Anchor index_rRegion Index.p2m Note that the region index is offset by 8, for instance, r009 is region 1 r015 is region 7, and r020 is region 12.

Naming Convention (testing dataset)

  • Notation in the .p2m file: **power.t001_Anchor index_r001.p2m Note that the region index is not shown in the name as each file corresponds to all the trajectory data (location is provided in x y z) in the .p2m file

File structure

Data
└─── _testing		
│   │  	HsMM_Experiment_power.t001_01.r001.p2m
│   │	...
│   │  	HsMM_Experiment_power.t001_01.r001.p2m
│
└─── _training
│   │  	HsMM_Experiment_power.t001_01.r009.p2m
│   │   ...
│   │  	HsMM_Experiment_power.t001_02.r009.p2m
│   │   ...
│   │  	HsMM_Experiment_power.t001_03.r009.p2m
│   │   ...
│   │   ...
│   │  	HsMM_Experiment_power.t001_08.r020.p2m


## Real Data Demo
The space is segmented into 21 uniform cells.
The black square shows the current true location of the user.
The rea shaded color indicates HsMM-FR estimation results.
The intensity of the red color scales according to its probability. If there are several cells being shaded,
it indicates that there is a disagreement among cells (ambiguity occures).


## Cite

@ARTICLE{HsMM-FR2023,
  author={Sun, Shuai and Li, Yan and Wang, Xuezhi and Moran, Bill and Rowe, Wayne S.T.},
  journal={IEEE Sensors Journal}, 
  title={Exploiting environmental information using HsMMs for smartphone user tracking}, 
  year={2023},
  volume={23},
  number={4},
  pages={4043-4051},
  doi={10.1109/JSEN.2023.3236642}

About

An illustrative example demonstrating how to apply the HsMM Forward Recursion algorithm for indoor user location region positioning.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages