DOA估计 - DML(deterministic ML) & SML(stochastic ML)

% Stochastic maximum likehood algorithm clear all; close all; clc; M = 8; % 阵元数 snapshots = 1024; % 信号长度 w = pi/4; % 信号频率 lambda = (2*pi*3e8)/w; % 信号波长 d = 0.5*lambda; % 阵元间距 snr = 5; % 信噪比 source_doa = 30; % 两个信号的入射角度 D = length(source_doa); % 信源数 A_SOI = [exp(-1j*(0:M-1)*d*2*pi*sind(source_doa)/lambda)].'; % 阵列流型 s = sqrt(10.^(snr/10))*wgn(snapshots,1,1,'complex')'; n = (1/sqrt(2))*(randn(M,snapshots)+1j*randn(M,snapshots)); x = A_SOI*s + n; % 含高斯白噪声接收信号 R_x = x*x'/snapshots; searching = -90:0.1:90; % 线阵的搜索范围为-90~90度 % for i = 1:length(searching) % A_theta = exp(-1j*(0:M-1)'*2*pi*d*sind(searching(i))/lambda); % % piA_theta = pinv(A_theta); % Pi_A_theta = A_theta*pinv(A_theta); % oPi_A_theta = eye(size(Pi_A_theta))-Pi_A_theta; % sigma_SML(i) = trace(oPi_A_theta)/(M-D); % P = pinv(A_theta)*(R-sigma_SML(i)*eye(M))*(A_theta); % P_SML(i) = log(det(A_theta*P*A_theta' + sigma_SML(i)*eye(M))); % end for i = 1:length(searching) A = exp(-1j*(0:M-1)'*2*pi*d*sind(searching(i))/lambda); A_plus = inv(A'*A)*A'; P_A = A*A_plus; Pc_A = eye(M) - P_A; sigma_SML(i) = trace(Pc_A*R_x)/(M-D); P_SML(i) = log(sigma_SML(i)^(M-D)*det(A_plus*R_x*A)); end % plot(searching,10*log10(P_SML./max(P_SML))); plot(searching,P_SML); for i = length(source_doa) xline(source_doa(i),'--'); end xlabel('theta/degree'); ylabel('P-SML/dB'); title('SML Algorithm for DOA'); grid on;