【无人机】具有自适应相移配置的UAV-RIS辅助系统的信道特征（Matlab代码实现）-CSDN博客

本文链接：https://blog.csdn.net/qq_46009436/article/details/144910991

💥💥💞💞欢迎来到本博客❤️❤️💥💥

🏆博主优势：🌞🌞🌞博客内容尽量做到思维缜密，逻辑清晰，为了方便读者。

⛳️座右铭：行百里者，半于九十。

📋📋📋本文目录如下：🎁🎁🎁

目录

⛳️赠与读者

💥1 概述

一、引言

二、系统模型

三、信道特征分析

四、研究方法与仿真分析

五、结论与展望

📚2 运行结果

🎉3 参考文献

🌈4 Matlab代码、文档

⛳️赠与读者

👨‍💻做科研，涉及到一个深在的思想系统，需要科研者逻辑缜密，踏实认真，但是不能只是努力，很多时候借力比努力更重要，然后还要有仰望星空的创新点和启发点。建议读者按目录次序逐一浏览，免得骤然跌入幽暗的迷宫找不到来时的路，它不足为你揭示全部问题的答案，但若能解答你胸中升起的一朵朵疑云，也未尝不会酿成晚霞斑斓的别一番景致，万一它给你带来了一场精神世界的苦雨，那就借机洗刷一下原来存放在那儿的“躺平”上的尘埃吧。

或许，雨过云收，神驰的天地更清朗.......🔎🔎🔎

💥1 概述

一、引言

随着无线通信技术的不断发展，无人机（UAV）和可重构智能表面（RIS）作为两种新兴技术，在提升无线通信系统性能方面展现出巨大潜力。UAV以其高机动性和灵活性，能够动态调整位置以优化通信链路；而RIS则通过智能地调整其表面元素的相移，实现对电磁波的精确调控。将UAV与RIS相结合，可以构建具有自适应相移配置的UAV-RIS辅助系统，进一步提升无线通信系统的信道性能和能效。

二、系统模型

在UAV-RIS辅助系统中，无人机搭载一个RIS，该RIS由大量可独立控制的反射元素组成，每个元素都能动态地调整其相移以优化信号传输。系统通常包括一个基站、一个或多个地面用户以及搭载RIS的无人机。基站通过RIS构建的级联信道向地面用户发送信号，而无人机则作为移动中继，通过调整其位置和RIS的相移配置来优化信道性能。

三、信道特征分析

路径损耗与阴影衰落：
- UAV-RIS辅助系统中的信道特性受到多种因素的影响，包括无人机的飞行高度、RIS的物理属性（如反射元素的尺寸、数量、排列方式等）以及通信环境的复杂性。
- 路径损耗和阴影衰落是信道特性的重要组成部分。路径损耗主要由信号在传播过程中的衰减引起，而阴影衰落则由于障碍物对信号的遮挡造成。
多径效应与时延扩展：
- 由于UAV和RIS的引入，系统中可能存在多条传播路径。这些路径的信号在到达接收端时会产生叠加，导致多径效应。
- 多径效应会引起信号的时延扩展，即信号在时间上被拉伸或压缩。时延扩展会影响信号的接收质量和系统的通信性能。
相移配置对信道特性的影响：
- RIS的相移配置对信道特性具有重要影响。通过智能地调整RIS的相移，可以优化信号的传输方向，减少信号间的干扰，提高信道的传输效率。
- 自适应相移配置算法可以根据通信环境的变化动态调整RIS的相移，以适应不同的通信需求。

四、研究方法与仿真分析

为了深入研究具有自适应相移配置的UAV-RIS辅助系统的信道特征，可以采用以下研究方法和仿真分析手段：

建立信道模型：
- 基于电磁波传播理论和无线通信原理，建立包含UAV、RIS和地面用户的信道模型。该模型应能够准确描述信号在传播过程中的衰减、反射、折射等现象。
仿真分析：
- 利用仿真软件对建立的信道模型进行仿真分析。通过调整无人机的飞行高度、RIS的相移配置等参数，观察信道特性的变化规律。
- 分析不同参数对信道性能的影响，如路径损耗、阴影衰落、多径效应和时延扩展等。
优化算法设计：
- 设计自适应相移配置算法，以优化UAV-RIS辅助系统的信道性能。算法应根据通信环境的变化动态调整RIS的相移配置。
- 对设计的优化算法进行仿真验证，评估其在实际应用中的性能和效果。

五、结论与展望

具有自适应相移配置的UAV-RIS辅助系统在提升无线通信系统性能方面展现出巨大潜力。通过深入研究其信道特征，可以为系统设计提供理论支撑和优化方向。未来，可以进一步探索更高效的自适应相移配置算法、更精确的信道建模方法以及更广泛的应用场景，以推动UAV-RIS辅助系统的实用化和商业化进程。

📚2 运行结果

部分代码：

clear all;
trials = 5e6;
%%
L = 1; % number of aerial CCI
K = 1; % number of ground CCI
M = 6; % number of transmit antenna
N = 12*12; % number of reflecting elements
%% Noise
B = 10*10^6; % Hz - Bandwidth
N0= -174; % dBm/Hz - Noise Spectral Density
AWGN = db2pow(N0)*B; % mW - Noise Power
%% 3D deployment
max_d = 1;
%
xS = 0;
yS = 0;
zS = 0;

% UAV
x_US = 0.5*max_d;
y_US = 0.5*max_d;
z_US = 1*max_d;

% DES
x_DS = 0.5*max_d;
y_DS = 0.5*max_d;
z_DS = 0;

% RIS
x_RS = 1.0*max_d;
y_RS = 1.0*max_d;
z_RS = 0*max_d;

% ACCI
if (L == 2)
x_IS= [0.2; 0.4]*max_d;
y_IS= [0.2; 0.8]*max_d;
z_IS= [0.6; 0.4]*max_d;
else
x_IS= 0.2*max_d;
y_IS= 0.2*max_d;
z_IS= 0.6*max_d;
end

% GCCI
if K == 2
x_JS= [0.4; 0.9]*max_d;
y_JS= [0.4; 0.9]*max_d;
z_JS= [0; 0]*max_d;
else
x_JS= 0.4*max_d;
y_JS= 0.4*max_d;
z_JS= 0*max_d;
end
%% Distance
d_SU = sqrt((xS-x_US).^2 + (yS-y_US).^2 + (zS-z_US).^2);
d_UR = sqrt((x_RS-x_US).^2 + (y_RS-y_US).^2 + (z_RS-z_US).^2);
d_UD = sqrt((x_DS-x_US).^2 + (y_DS-y_US).^2 + (z_DS-z_US).^2);
d_IU = sqrt((x_IS-x_US).^2 + (y_IS-y_US).^2 + (z_IS-z_US).^2);
d_JU = sqrt((x_JS-x_US).^2 + (y_JS-y_US).^2 + (z_JS-z_US).^2);
d_ID = sqrt((x_IS-x_DS).^2 + (y_IS-y_DS).^2 + (z_IS-z_DS).^2);
d_JD = sqrt((x_JS-x_DS).^2 + (y_JS-y_DS).^2 + (z_JS-z_DS).^2);
d_RD = sqrt((x_RS-x_DS).^2 + (y_RS-y_DS).^2 + (z_RS-z_DS).^2);
%% Large-Scale Fading (Pathloss)
Gt = db2pow(0); % 0 - dBi
Gr = db2pow(0); % 0 - dBi
fc = 3; % GHz
PL = @(d) pathloss3GPP_UMi(Gr,Gt,fc,d);
PL_SU = PL(d_SU);
PL_UR = PL(d_UR);
PL_UD = PL(d_UD);
PL_IU = PL(d_IU);
PL_JU = PL(d_JU);
PL_ID = PL(d_ID);
PL_JD = PL(d_JD);
PL_RD = db2pow(- Gr - Gt - 37.3 + 26*log10(fc) - 36.7*log10(d_RD));
%% Small-Scale Fading Constants
K0 = 10^(0/10);
Kpi= 10^(5/10);
K1 = K0;
K2 = 2/pi*log(Kpi/K0);

kappa_ab = @(za,zb,dab) asin(abs(za-zb)./dab); % works for elevation in [-pi/2, pi/2]
kappa_SU = kappa_ab(zS,z_US,d_SU);
kappa_UR = kappa_ab(z_RS, z_US, d_UR);
kappa_UD = kappa_ab(z_DS, z_US, d_UD);
kappa_IU = kappa_ab(z_IS, z_US, d_IU);
kappa_JU = kappa_ab(z_JS, z_US, d_JU);
kappa_ID = kappa_ab(z_IS, z_DS, d_ID);
kappa_JD = kappa_ab(z_JS, z_DS, d_JD);
kappa_RD = kappa_ab(z_RS, z_DS, d_RD);

K_SU = K1*exp(K2*kappa_SU); % Rician factor of gamma_SU
K_UR = K1*exp(K2*kappa_UR); % Rician factor of gamma_UR
K_UD = K1*exp(K2*kappa_UD); % Rician factor of gamma_UD
K_IU = K1*exp(K2*kappa_IU); % Rician factor of gamma_IU
K_JU = K1*exp(K2*kappa_JU); % Rician factor of gamma_JU
K_ID = K1*exp(K2*kappa_ID); % Rician factor of gamma_ID
K_JD = K1*exp(K2*kappa_JD); % Rician factor of gamma_JD
K_RD = K1*exp(K2*kappa_RD); % Rician factor of gamma_RD

lambda_SU = (K_SU + 1)/PL_SU;
lambda_UR = (K_UR + 1)/PL_UR;
lambda_UD = (K_UD + 1)/PL_UD;
lambda_IU = (K_IU + 1)/PL_IU;
lambda_JU = (K_JU + 1)/PL_JU;
lambda_ID = (K_ID + 1)/PL_ID;
lambda_JD = (K_JD + 1)/PL_JD;
lambda_RD = (K_RD + 1)/PL_RD;
%% Normalized Cascaded Gain & Normalized SNR (2st hop)
filename = sprintf('gamma_cascaded_N%d.mat', N);
try
load(filename);
catch
gamma_cascaded = 0;
for n = 1:N
%
gamma_UR = ncx2rnd(2, 2*K_UR, [1,trials])/(2*(K_UR+1));
gamma_RD = ncx2rnd(2, 2*K_RD, [1,trials])/(2*(K_RD+1));
%
gamma_cascaded = gamma_cascaded + sqrt(gamma_UR.*gamma_RD);
end
gamma_cascaded = gamma_cascaded.^2;
%
save(filename, 'gamma_cascaded');
end
%
%% Normalized A2G SNR
ibeta = 1;

generate_marcumqTable(K_UD);
marcumqTable = cell2mat(struct2cell(load('marcumqTable.mat')));
inx = (1:9:100);
marcumqTable = marcumqTable(inx, :);
betas = marcumqTable(:,1);
% 0
% 0.010067047333128
% 0.059757157288055
% 0.191426537753562
% 0.416100628846570
% 0.669431828304073
% 0.860516311805993
% 0.957572629260874
% 0.990891287023915
% 0.998638110850211
% 0.999859415115188
% 0.999990038049460

beta = betas(ibeta);
tau = marcumqTable(ibeta, 2)^2 / (2*lambda_UD);

fprintf('beta = %f, tau = %f dB \n', beta, pow2db(tau));

filename = sprintf('gamma_UD_%d.mat', ibeta);
try
load(filename);
catch
P = [beta 1-beta; beta 1-beta];
mc = dtmc(P); S = simulate(mc, trials-1).'-1;
% ---------------------------------------------------------------------
% Execute for new nodes' locations or K factor
% ---------------------------------------------------------------------
gamma_UD = [];

[S_sort, I_sort] = sort(S, 'descend');
no_state_1 = find(S_sort, 1, 'last');

% Sampling channel with state 1
% ---------------------------------------------------------------------
filename = sprintf('gamma_UD_%d.mat', ibeta);
try
load(filename);
catch
while(1)
sample = ncx2rnd(2, 2*K_UD, [1, trials])/(2*(K_UD+1));

good_samples = find(sample > tau/PL_UD);
if ~isempty(good_samples)
gamma_UD = [gamma_UD, sample(good_samples)];
end
if length(gamma_UD) >= no_state_1
gamma_UD = gamma_UD(1:no_state_1); break;
end
end
gamma_UD = [gamma_UD zeros(1, trials-no_state_1)];

% De-sorting
gamma_UD = gamma_UD(I_sort);

filename = sprintf('gamma_UD_%d.mat', ibeta);
save(filename, 'gamma_UD');
end
end

rlfCoeff = db2pow(0);
bg_cascaded = (rlfCoeff^2*PL_UR*PL_RD);
gamma_t = PL_UD*gamma_UD/bg_cascaded;

gamma_a2g = sqrt(gamma_cascaded) + sqrt(gamma_t);
gamma_a2g = gamma_a2g.^2;
%
%% Normalized G2A SNR
gamma_g2a = sum(ncx2rnd(2, 2*K_SU, [M, trials])/(2*(K_SU+1)), 1);
%% Aerial Interference Power
PI = db2pow(0*ones(L,1));
bg_IU = PI/AWGN.*PL_IU;
bg_ID = PI/AWGN.*PL_ID;
gamma_IU = 0;
gamma_ID = 0;
for ii = 1:L
%
gamma_IU = gamma_IU + bg_IU(ii)*ncx2rnd(2, 2*K_IU(ii), [1, trials])/(2*(K_IU(ii)+1));
gamma_ID = gamma_ID + bg_ID(ii)*ncx2rnd(2, 2*K_ID(ii), [1, trials])/(2*(K_ID(ii)+1));
end
%% Ground Interference Power
PJ = db2pow(0*ones(K,1));
bg_JU = PJ/AWGN.*PL_JU;
bg_JD = PJ/AWGN.*PL_JD;
gamma_JU = 0;
gamma_JD = 0;
for ii = 1:K
%
gamma_JU = gamma_JU + bg_JU(ii)*ncx2rnd(2, 2*K_JU(ii), [1, trials])/(2*(K_JU(ii)+1));
gamma_JD = gamma_JD + bg_JD(ii)*ncx2rnd(2, 2*K_JD(ii), [1, trials])/(2*(K_JD(ii)+1));
end
%% Normalized SINR G2A
bg_g2a = 1/AWGN*PL_SU;
sinr_g2a = bg_g2a*gamma_g2a./(gamma_IU + gamma_JU + 1);
%% Normalized SINR A2G
sinr_a2g = 1/AWGN*bg_cascaded*gamma_a2g./(gamma_ID + gamma_JD + 1);
%% PDF & CDF of gamma_SU
m_g2a = 4;
N_g2a = M;

mu_gamma_XY = @(k, K) (1+K).^(-k).*gamma(1+k).*laguerreL(k, -K);
mu_g2a_1 = M*mu_gamma_XY(1, K_SU);
mu_g2a_2 = M*mu_gamma_XY(2, K_SU) + (M-1)/M*mu_g2a_1^2;

disp('Matching G2A SNR to SSR:...')
[param_g2a, pdf_gamma_g2a, cdf_gamma_g2a] = fitdist_SRS(mu_g2a_1, mu_g2a_2, m_g2a, N_g2a);

xx = logspace(pow2db(min(gamma_g2a))/10, pow2db(max(gamma_g2a))/10, 25);
data_cdf_snr_g2a = zeros(size(xx));
for ix = 1:length(xx)
data_cdf_snr_g2a(ix) = cdf_gamma_g2a(xx(ix));
end

figure;
ecdf(gamma_g2a); hold on;
plot(xx, data_cdf_snr_g2a); hold off;
axis([min(xx) max(xx) 0 1]);
set(gca, 'XScale', 'Log');
set(gca, 'YScale', 'Log');

filename = sprintf('gamma_SU_M%d.mat', M);
save(filename, 'gamma_g2a');
filename = sprintf('y_cdf_snr_g2a_M%d.mat', M);
save(filename, 'data_cdf_snr_g2a');
filename = sprintf('x_cdf_snr_g2a_M%d.mat', M);
save(filename, 'xx');
%% PDF & CDF of G2A SINR
kappa_g2a = param_g2a(1);
omega_g2a = param_g2a(2);

disp('Calculating G2A SINR...')
cdf_sinr_g2a = @(x) cdf_sinr(N_g2a, m_g2a, kappa_g2a, omega_g2a, K_IU, bg_IU, K_JU, bg_JU, x/bg_g2a);

xx = logspace(pow2db(min(sinr_g2a))/10, pow2db(max(sinr_g2a))/10, 25);
data_cdf_sinr_g2a = zeros(size(xx));
for ix = 1:length(xx)
data_cdf_sinr_g2a(ix) = cdf_sinr_g2a(xx(ix));
end
figure;
ecdf(sinr_g2a); hold on;
plot(xx, data_cdf_sinr_g2a); hold off;
axis([min(xx) max(xx) 0 1]);
set(gca, 'XScale', 'Log');
set(gca, 'YScale', 'Log');

filename = sprintf('sinr_g2a_M%d.mat', M);
save(filename, 'sinr_g2a');
filename = sprintf('y_cdf_sinr_g2a_M%d.mat', M);
save(filename, 'data_cdf_sinr_g2a');
filename = sprintf('x_cdf_sinr_g2a_M%d.mat', M);
save(filename, 'xx');
%% PDF & CDF of A2G SNR
f_gamma_UD = @(x) exp(-lambda_UD*x-K_UD).*lambda_UD.*besseli(0, 2*sqrt(K_UD*lambda_UD*x));
Infinity = 2*1e6;
F_gamma_UD = @(x) 1 - marcumq(sqrt(2*K_UD), sqrt(2*lambda_UD*x));

% Transition Matrix
P = [beta 1-beta; beta 1-beta];
% Steady-State Probabilities
pi1 = P(1,2)/(P(1,2)+P(2,1));
pi0 = P(2,1)/(P(1,2)+P(2,1));

% kth moment of sqrt(gamma_cascaded)
mu_cascaded = @(k) kmoment_gamma_cascaded(N, k, K_UR, K_RD);

% kth moment of sqrt(gamma_direct)
mu_direct = @(k) pi1*(bg_cascaded)^(-k/2)/(1-beta)...
* integral(@(t) t.^(k/2) .* f_gamma_UD(t), tau, Infinity);

mu_gamma_a2g = @(k) kmoment_gamma_a2g(k, mu_cascaded, mu_direct);
%
m_a2g = 2; Q_a2g = 20;
disp('Matching A2G SNR to SSR:...')
[param_a2g, ~, cdf_snr_a2g] = fitdist_SRS(mu_gamma_a2g(1), mu_gamma_a2g(2), m_a2g, Q_a2g);

xx = logspace(pow2db(min(gamma_a2g))/10, pow2db(max(gamma_a2g))/10, 25);
data_cdf_snr_a2g = zeros(size(xx));
for ix = 1:length(xx)
data_cdf_snr_a2g(ix) = cdf_snr_a2g(xx(ix));
end

figure;
ecdf(gamma_a2g); hold on;
plot(xx, data_cdf_snr_a2g); hold off;
axis([min(xx) max(xx) 0 1]);

filename = sprintf('gamma_a2g_N%d.mat', N);
save(filename, 'gamma_a2g');
filename = sprintf('y_cdf_snr_a2g_N%d.mat', N);
save(filename, 'data_cdf_snr_a2g');
filename = sprintf('x_cdf_snr_a2g_N%d.mat', N);
save(filename, 'xx');
%% PDF & CDF of A2G SINR
kappa_a2g = param_a2g(1);
omega_a2g = param_a2g(2);
Q_a2g = param_a2g(6);
bg_a2g = (1/AWGN)*bg_cascaded;

disp('Calculating A2G SINR...')
cdf_sinr_a2g = @(x) cdf_sinr(Q_a2g, m_a2g, kappa_a2g, omega_a2g, K_ID, bg_ID, K_JD, bg_JD, x/bg_a2g);

xx = logspace(pow2db(min(sinr_a2g))/10, pow2db(max(sinr_a2g))/10, 25);
data_cdf_sinr_a2g = zeros(size(xx));
for ix = 1:length(xx)
data_cdf_sinr_a2g(ix) = cdf_sinr_a2g(xx(ix));
end
figure;
ecdf(sinr_a2g); hold on;
plot(xx, data_cdf_sinr_a2g); hold off;
axis([min(xx) max(xx) 0 1]);

filename = sprintf('sinr_a2g_N%d.mat', N);
save(filename, 'sinr_a2g');
filename = sprintf('y_cdf_sinr_a2g_N%d.mat', N);
save(filename, 'data_cdf_sinr_a2g');
filename = sprintf('x_cdf_sinr_a2g_N%d.mat', N);
save(filename, 'xx');
%% PDF & CDF of e2e SINR
kappa_a2g = param_a2g(1);
omega_a2g = param_a2g(2);
bg_a2g = (1/AWGN)*bg_cascaded;

disp('Calculating e2e SINR...')
cdf_sinr_e2e = @(x) 1 - (1-cdf_sinr_a2g(x))*(1-cdf_sinr_g2a(x));
sinr_e2e = min([sinr_g2a; sinr_a2g]);

xx = logspace(pow2db(min(sinr_e2e))/10, pow2db(max(sinr_e2e))/10, 50);
data_cdf_sinr_e2e = zeros(size(xx));
for ix = 1:length(xx)
data_cdf_sinr_e2e(ix) = cdf_sinr_e2e(xx(ix));
end
figure;
ecdf(sinr_e2e); hold on;
plot(xx, data_cdf_sinr_e2e); hold off;
axis([min(xx) max(xx) 0 1]);
set(gca, 'XScale', 'Log');
set(gca, 'YScale', 'Log');

filename = sprintf('sinr_e2e_M%d_N%d.mat', M, N);
save(filename, 'sinr_e2e');
filename = sprintf('y_cdf_sinr_e2e_M%d_N%d.mat', M, N);
save(filename, 'data_cdf_sinr_e2e');
filename = sprintf('x_cdf_sinr_e2e_M%d_N%d.mat', M, N);
save(filename, 'xx');

🎉3 参考文献

文章中一些内容引自网络，会注明出处或引用为参考文献，难免有未尽之处，如有不妥，请随时联系删除。(文章内容仅供参考，具体效果以运行结果为准)

[1] Michailidis E T , Miridakis N I , Michalas A ,et al.Energy Optimization in Dual-RIS UAV-Aided MEC-Enabled Internet of Vehicles[J].Sensors, 2021(13).DOI:10.3390/S21134392.

[2]黄星博.无人机携带RIS辅助的中继协作通信系统研究[D].南京邮电大学,2023.

[3]郭雅婧,孙强,章嘉懿,等.基于智能超表面的无人机通信联合波束赋形与航迹规划[J].南通大学学报:自然科学版, 2021, 20(3):7.

[4]胡浪涛,毕松姣,刘全金,等.基于强化学习的智能超表面辅助无人机通信系统物理层安全算法[J].电子与信息学报, 2022, 44(7):9.DOI:10.11999/JEIT211613.