【WSN】基于WSN下的多目标跟踪定位附matlab代码
✅作者简介：热爱科研的Matlab仿真开发者，修心和技术同步精进，matlab项目合作可私信。
个人主页：Matlab科研工作室
个人信条：格物致知。
更多Matlab仿真内容点击
智能优化算法 神经网络预测 雷达通信 无线传感器
信号处理 图像处理 路径规划 元胞自动机 无人机
⛄ 内容介绍
近年来,无线传感器网络以其低成本、低功耗、布设方便等特点,普遍应用于军事、民用领域,诸如监控系统、导航系统、障碍规避系统等。这一类系统共同要解决的问题是如何确定目标的位置,以及如何确定目标的运动方向或轨迹等。因此,在无线传感器网络环境下实现一种有效、快速的定位算法已经成为当今的一个研究热点。
⛄ 部分代码
% Executable
clc, clear all, close all
tic
profile on
% Initialization
users = 4;
dimensions = [-12 12; -12 12]; % [x_min x_max; y_min y_max]
total_steps = 75;
calibration_steps = 10;
precision = 0.5;
mean_distance = 6; % Mean distance in positions
step_distance = 0.75; % in meters
%% Input
% Paths that the targets follow inside the considered map
users_path = create_path(users, dimensions, total_steps, step_distance, mean_distance);
% Plot user path
figure('name','Users paths')
for i = 1:users
plot(users_path(1,:,i),users_path(2,:,i))
hold on
end
hold on
grid on
title('Path of each user in the map')
xlabel('X')
ylabel('Y')
axis([dimensions(1,1) dimensions(1,2) dimensions(2,1) dimensions(2,2)])
%% Generalized radiation
% Footprint on the map caused by the presence of a target
% Gaussian radiation characteristics
radiation_amplitude = 20; % Maximum change in signal, when a target is present
target_width = [1 0.5]; % Target width (meters) in dimensions [x y]
rotation = 0; % Target rotation in radians
noiselevel = 0.25; % Noise deviation for the gaussian function (i.e. noiselevel*randn)
% RF radiation
radiation = create_radiation(dimensions, users_path, precision,calibration_steps, radiation_amplitude, target_width, rotation, noiselevel );
%% RF section
% Measurements based on the change in RSS (received signal strength) in
% each link (i.e. conexion between two sensors) caused by each voxel
% RF model - Change in RSS by each voxel - Simulation
rss_change = radiation;
% RF links - Change in RSS by each link - Simulation
sensor_position = [-12 -6 0 6 12 -12 -6 0 6 12 -12 -6 0 6 12 -12 -6 0 6 12 -12 -6 0 6 12; -12 -12 -12 -12 -12 -6 -6 -6 -6 -6 0 0 0 0 0 6 6 6 6 6 12 12 12 12 12];
%sensor_position = [dimensions(1,1) dimensions(1,2)/2 0 0 ; dimensions(2,1) 0 dimensions(2,1)/2 dimensions(2,2)/2];
nsensors = length(sensor_position(1,:));
nlinks = (nsensors^2-nsensors)/2; % Number of links: L = (K^2-K)/2 (K: number of sensors)
link_weights = locate_link_ellipses(dimensions,sensor_position,precision,nlinks);
rss_change_link = rss_links(rss_change,link_weights,nlinks);
%% RTI Image Estimation
% This section is part of the estimation algorithm, after the simulation
% (or real environment test)
% Using Regularized Least-Squares (RLS) approach, the RSS change in each
% voxel is estimated (in simulation this is called "rss_change")
rss_change_estimate = rss_estimation(dimensions, rss_change_link, link_weights, precision);
%plottracking(1,radiation,users_path,dimensions, calibration_steps, 2);
% Gaussian filter (not implemented yet, excessive execution time)
%% Multiple Target Tracking with RTI images (RF sensors)
global istarget;
istarget = false; % Empty area or monitoring targets
% Initialization of variables
filtered_rti = zeros(size(rss_change_estimate(:,:,calibration_steps+1:end)));
alfa_f = 0.9;
beta = 0.8;
xaxis = (dimensions(1,1)+precision/2):precision:(dimensions(1,2)-precision/2);
yaxis = (dimensions(2,1)+precision/2):precision:(dimensions(2,2)-precision/2);
total_clusters = cell(1,total_steps);
% Threshold when the monitored area is empty
Ie = mean(max(max(rss_change_estimate(:,:,1:calibration_steps)))); % Average maximum intensity in the training period
Tt = [2*Ie 0]; % Initial Detection Threshold
% Initizalization of Threshold when targets are being tracked
I_lpf = ones(1,total_steps);
% Tracking algorithm initialization
% Parameters
Re_width = 2; % Entrance/Exit region width in meters
radius_t = 2; % Initial radius around a target in meters [m]
n_memory = 3; % Number of previous decisions (clusters) taken into account for instant k
n_confirm = 5; % To confirm a track, it has to be assigned to the same ID at least 'n_confirm' times in the last 'n_max_confirm' observations
n_max_confirm = 10;
n_delete = 5; % To delete a track, it has to be not assigned in the last 'n_delete' observations
% Initializing tracking variables:
targets = cell(1,total_steps);
target_id = 1;
confirmed_tracks = []; % Confirmed tracks after N of M observations
track_candidates = []; % Track candidates to be confirmed or deleted
deleted_tracks = []; % Deleted track candidates if not confirmed
figure(2)
hold on
subplot(2,3,1) % Subplot 1 -- Users Paths
for i = 1:users
plot(users_path(1,:,i),users_path(2,:,i))
hold on
end
grid on
title('Users path')
xlabel('X')
ylabel('Y')
axis([dimensions(1,1) dimensions(1,2) dimensions(2,1) dimensions(2,2)])
for step = 1:total_steps
% Thresholding
filtered_rti(:,:,step) = detection_thresholding(rss_change_estimate(:,:,calibration_steps+step),Tt);
% Clustering
[clusters_rti, total_clusters{1,step}, voxels_forcluster] = clustering(filtered_rti(:,:,step), precision);
% Tracking
[targets, track_candidates, confirmed_tracks, deleted_tracks, target_id ] = target_tracking( targets, total_clusters, step,...
dimensions, precision, Re_width, radius_t, n_memory, n_confirm, n_max_confirm, n_delete, target_id, track_candidates, ...
confirmed_tracks, deleted_tracks );
if istarget
% Iterative thresholding
Imin_previous = zeros(1,length(voxels_forcluster(:,1)));
for pos = 1:length(voxels_forcluster(:,1))
Imin_previous(pos) = filtered_rti(voxels_forcluster(pos,1),voxels_forcluster(pos,2),step);
end
Imin = min(Imin_previous);
I_lpf(step+1) = alfa_f*I_lpf(step)+(1-alfa_f)*Imin; % Low-pass filter of Imin
Tt(2) = beta*I_lpf(step+1);
%% Plotting
rows_inmeters = (voxels_forcluster(:,1).*precision+dimensions(1,1))-precision/2;
columns_inmeters = (voxels_forcluster(:,2).*precision+dimensions(2,1))-precision/2;
subplot(2,3,2) % Subplot 2 -- Input radiation
surf(xaxis,yaxis,rss_change(:,:,calibration_steps+step)','EdgeColor','none')
view(2);
colormap jet;
grid on;
caxis([0 max(max(rss_change(:,:,calibration_steps+step)))]);
grid on
title('Input Radiation')
xlabel('X')
ylabel('Y')
axis([dimensions(1,1) dimensions(1,2) dimensions(2,1) dimensions(2,2)])
subplot(2,3,3) % Subplot 3 -- Estimated radiation from sensors
surf(xaxis,yaxis,rss_change_estimate(:,:,calibration_steps+step)','EdgeColor','none')
view(2);
colormap jet;
grid on;
caxis([0 max(max(rss_change_estimate(:,:,calibration_steps+step)))]);
grid on
title('Estimated Radiation')
xlabel('X')
ylabel('Y')
axis([dimensions(1,1) dimensions(1,2) dimensions(2,1) dimensions(2,2)])
subplot(2,3,4) % Subplot 4 -- Filtered estimated radiation
surf(xaxis,yaxis,filtered_rti(:,:,step)','EdgeColor','none')
view(2)
caxis([0 max(max(filtered_rti(:,:,step)))]);
grid on;
axis([dimensions(1,1) dimensions(1,2) dimensions(2,1) dimensions(2,2)])
xlabel('[m]');
ylabel('[m]');
title('Filtered estimated RTI')
subplot(2,3,5) % Subplot 5 -- Clusters
scatter(rows_inmeters,columns_inmeters,10,clusters_rti)
grid on;
axis([dimensions(1,1) dimensions(1,2) dimensions(2,1) dimensions(2,2)])
xlabel('[m]');
ylabel('[m]');
title('Clusters in estimated RTI')
pause(0.1)
subplot(2,3,6) % Subplot 6 -- Multi target Tracking
hold on
for b = 1:length(targets(:,step))
if isstruct(targets{b,step})
target_x_position_inmeters = ((targets{b,step}.position(1)).*precision+dimensions(1,1))-precision/2;
target_y_position_inmeters = ((targets{b,step}.position(2)).*precision+dimensions(1,1))-precision/2;
plot(target_x_position_inmeters,target_y_position_inmeters,'w*')
hold on
ln = findobj('type','line');
set(ln, 'marker', '.', 'markers', 10, 'markerfa', 'w')
text(target_x_position_inmeters,target_y_position_inmeters,num2str(targets{b,step}.ID))
grid on;
axis([dimensions(1,1) dimensions(1,2) dimensions(2,1) dimensions(2,2)])
xlabel('[m]');
ylabel('[m]');
title('Multi Target Tracking')
pause(0.1)
end
end
subplot(2,3,2)
cla
subplot(2,3,3)
cla
subplot(2,3,4)
cla
subplot(2,3,5)
cla
else
subplot(2,3,2) % Subplot 2 -- Input radiation
surf(xaxis,yaxis,rss_change(:,:,calibration_steps+step)','EdgeColor','none')
view(2);
colormap jet;
grid on;
caxis([0 max(max(rss_change(:,:,calibration_steps+step)))]);
grid on
title('Input Radiation')
xlabel('X')
ylabel('Y')
axis([dimensions(1,1) dimensions(1,2) dimensions(2,1) dimensions(2,2)])
subplot(2,3,3) % Subplot 3 -- Estimated radiation from sensors
surf(xaxis,yaxis,rss_change_estimate(:,:,calibration_steps+step)','EdgeColor','none')
view(2);
colormap jet;
grid on;
caxis([0 max(max(rss_change_estimate(:,:,calibration_steps+step)))]);
grid on
title('Estimated Radiation')
xlabel('X')
ylabel('Y')
axis([dimensions(1,1) dimensions(1,2) dimensions(2,1) dimensions(2,2)])
subplot(2,3,4) % Subplot 4 -- Filtered radiation
surf(xaxis,yaxis,filtered_rti(:,:,step)','EdgeColor','none')
view(2)
caxis([0 max(max(filtered_rti(:,:,step)))]);
grid on;
axis([dimensions(1,1) dimensions(1,2) dimensions(2,1) dimensions(2,2)])
xlabel('[m]');
ylabel('[m]');
title('Filtered estimated RTI')
subplot(2,3,5) % Subplot 5 -- Clusters
plot(0,0);
grid on;
axis([dimensions(1,1) dimensions(1,2) dimensions(2,1) dimensions(2,2)])
xlabel('[m]');
ylabel('[m]');
title('Clusters in estimated RTI')
pause(0.1)
subplot(2,3,6) % Subplot 6 -- Multi target tracking
plot(0,0);
hold on
grid on;
axis([dimensions(1,1) dimensions(1,2) dimensions(2,1) dimensions(2,2)])
xlabel('[m]');
ylabel('[m]');
title('Multi target tracking')
pause(0.1)
subplot(2,3,2)
cla
subplot(2,3,3)
cla
subplot(2,3,4)
cla
subplot(2,3,5)
cla
end
end
% Image intensity normalization
normalized_rti = rti_normalization(filtered_rti);
%% Target tracking algorithm
% Tracking
%users_track = users_path + randn(2,total_steps,users);
% Plot radiation for all users
% loops = 1;
% fps = 12;
% nview = 2;
%plottracking(radiation,users_path, users_track, dimensions, calibration_steps, nview);
% clip2 = plottracking(rss_change_estimate,users_path, users_track, dimensions, calibration_steps,nview);
% clip3 = plottracking(filtered_rti,users_path, users_track, dimensions, calibration_steps,nview);
% movie(clip,loops,fps);
profile viewer
toc
⛄ 运行结果
⛄ 参考文献
[1]沙勇. "基于Matlab的WSN定位算法仿真设计." 齐齐哈尔大学学报：自然科学版 33.6(2017):3.
[2]叶景志, 赵玲, 罗汶锋,等. 基于WSN的多目标跟踪及反馈控制平台[J]. 计算机工程, 2012, 38(10):77-79.
❤️ 关注我领取海量matlab电子书和数学建模资料
❤️部分理论引用网络文献，若有侵权联系博主删除