This is the MATLAB implementation of the work presented in RSS-Based Localization in WSNs Using Gaussian Mixture Model via Semidefinite Relaxation.
pathLossModel.m : Plot the path loss model and the histogram of the Gaussian Mixture Model
estimatePos.m : Returns the estimated target position using SDP in CVX
export_CDF_GM_SDP.m : Creates matrix sdpCDF.mat containing CDF for GM-SDP-2
export_CDF_WLS.m : Creates matrix wlsCDF.mat containing CDF for weighted least square (WLS)
export_crlb.m : Creates matrix crlb.mat containing Cramer-Rao Lower Bound (CRLB) for WSN Localization
export_GM_SDP.m : Creates matrix SDPrmse.mat containing RMSE for GM-SDP-2
export_WLS.m : Creates matrix SDPrmse.mat containing RMSE for WLS
findCrlb.m : Returns CRLB for a particular target and anchor placement
findRSS.m : Returns the Received Signal Strength (RSS) at all target nodes in a WSN
monteCarloInt.m : Returns the value of monte-carlo integration used in calculating the fisher information matrix
place.m : Used for setting the location of target and anchor nodes in WSN
plot_CDF.m : Used for plotting the CDF of various localization algorithms from their .mat files
plot_RMSE.m : Used for plotting the RMSE of various localization algorithms from their .mat files
Saved output folder contains .mat files of the variables plotted in the result screenshots section.
CVX
Statistics and Machine Learning Toolbox
[1] WSN
| Example WSN |
|---|
 |
[2] Path Loss Model
| Path Loss Model |
|---|
 |
[3] RMSE and CDF
| RMSE v/s N (number of anchors) | CDF v/s error |
|---|---|
 |
 |
Note: Slightly different anchor placement was used in the WSN localization simulation.
[1] Edit place.m for changing target and anchor node location.
[2] Run export_GM_SDP.m, export_WLS.m, and export_crlb.m to generate .mat files for RMSE.
[3] Run plot_RMSE.m to plot RMSE vs N.
[4] Run export_CDF_GM_SDP.m, and export_CDF_WLS.m to generate .mat files for CDF.
[5] Run plot_CDF.m to plot CDF vs error.
[1] Zhang, Yueyue, et al. "RSS-based localization in WSNs using Gaussian mixture model via semidefinite relaxation." IEEE Communications Letters 21.6 (2017): 1329-1332.
[2] http://cvxr.com/cvx/