Name		Name	Last commit message	Last commit date
Latest commit History 6 Commits
aoa		aoa
data		data
figures		figures
log		log
model		model
README.md		README.md
ble-based-smartphone-localization.pdf		ble-based-smartphone-localization.pdf
check_package.py		check_package.py
deployment.py		deployment.py
imu.py		imu.py
kalman_sim.py		kalman_sim.py
plot_figures.py		plot_figures.py
train.py		train.py
utils.py		utils.py

Repository files navigation

Directory Tree

check_package.py: check a list of Python packages which are needed for the environment
utils.py: basic utility functions which are used by other scripts
train.py: compare different machine learning algorithms(Logistic, LDA, QDA, KNN, SVM, RF, GBM, MLP)
deployment.py: generate Java models using h2o package
kalman_sim.py: illustrate the increased performance for object tracking using Kalman filters(CA or CV model)
imu.py: attempt to use IMU values of smartphone to improve the localization performance(Smartphone and PC need to be in the same network in order to receive the transmitted data by the smartphone)
plot_figures.py: plot analytical figures which are also used by the following report
ble-based-smartphone-localization.pdf: localization report which includes challenge issues and the detailed algorithm
aoa
- aoa.py: connect to NXP board and show real time AOA analysis
- aoa_sim.py: simulate the AOA results using different antenna arrays(linear, square, circular)
figures: generated analytical figures which are then used by the localization report
data: intermediate data files generated by the execution of script
model: generated Java model files which are then used by Android
- h2o-genmodel.jar: h2o jar to process generated Jave models
- EightNormal.java: Java model for zone prediction
- MLP4Px.java: Java model for coordinate x prediction
- MLP4Py.java: Java model for coordinate y prediction
log: RSSI log files for different purposes(training or analysis)
- circular antenna: measurements using circular antenna
  - anechonic chamber: measurements taken in the anechonic chamber
  - around car: measurements taken in the outside open space around the vehicle
- heatmap: measurements of every point around the vehicle which were taken line by line
- influence: analysis of encountered challenge
  - body: present the body obstacle attenuation
  - channel: present the difference of RSSI values among each channel(37, 38, 39)
  - enviroment: present the difference of RSSI values between inside building and outside building
  - smartphone: present the existence of offset among different smartphone models
  - wifi: present the potential influence of wifi when the wifi of the smartphone is activated
- path loss: experimental path loss
  - Circular: path loss of the circular antenna
  - PIFA: path loss of the PIFA antenna
- peps mini: 3 classes(access, internal, lock) classification with normal usage purpose
  - access: train set taken around the vehicle in less than 2m
  - internal: train set taken inside of the vehicle including trunk zone
  - lock:: train set taken outside the vehicle in more than 2.3m
- peps normal: 7 classes(front, left, right, back, start, trunk, lock) classification with normal usage purpose
  - front/left/right/back: train set taken in each side of the vehicle and in less than 2m
  - start: train set taken inside of the vehicle(seat zone)
  - trunk: train set taken in the trunk zone
  - lock: train set taken in the far zone in more than 2.3m
- peps thatcham: 7 classes(front, left, right, back, start, trunk, lock) classification with thatcham purpose
  - front/left/right/back: train set taken in each side of the vehicle and in less than 1.5m
  - start: train set taken inside of the vehicle(seat zone)
  - trunk: train set taken in the trunk zone
  - lock: train set taken in the far zone in more than 1.8m

Measurement suggestion

1. Zone prediction

walk slowly in each zone with the smartphone in the hand and in the face of the vehicle for 3-4 minutes to obtain a log file
a number of log files for each zone are needed(>=4 is preferred), this can be done either by demanding different persons
or taking the measurements at different time by the same person
It's better to have different walking paths when constructing data sets for the same zone
internal measurements can be done by putting the smartphone in some position for a certain time(~5s) and then moving to another position

2. Coordinate prediction

the distance between each point is 0.5m, the total measurement size is 11*10m
an equipment is preferred in order to stabilize the smartphone when taking the measurement
put the smartphone in the equipment and measure for a certain time(10-15s), and then move to another point
there is a tool in the smartphone IHM in order to memorize the number of the different point

About

No description, website, or topics provided.

Report repository

Releases

No releases published

Packages

Contributors

Languages

Java 97.5%
Python 2.5%