energydetection_k秩准则_K._or_k秩_秩K

clear all; clc; n =10000; %采样点数 pfa = 0.01; %虚警概率 snr1=-25:1:0; %信噪比 k = length(snr1); for i = 1:k %不同信噪比下仿真 s = 0; %1个数初值设为0，每次循环都清零 snr(i) = power(10,snr1(i)/10);%db转化十进制 for kk = 1:1 %每种信噪比下仿真次数 t = 1/n:1/n:1; %离散点1000 x = sin(pi*t); x = ( x > 0); %大于0判为1，否则0，方波产生方法 x =2*x-1; %信源数据产生 noise = randn(1,length(t)); %均值0方差1的噪声 pn = (std(noise)).^2; %噪声功率 a = sqrt(2*snr(i)); %信号幅值 xx = a.*sin(2*pi*4*t); %调制信号 y = x.*xx; %bpsk调制信号 ps = mean(abs(y).^2); %信号平均功率 yy = y+noise; %输出信号 r = (n+sqrt(2*n)*sqrt(2)*erfcinv(2*pfa)); %r = gaminv(1-paf,n/2,1)*2; sum1 = sum(abs(yy).^2); %输出信号的累积功率 if sum1>r s = s+1; %输出大于阈值及判为1，否则0，统计1的个数 end end p(i) = s/100; %检测概率：1的个数/n pd(i) = 0.5*erfc((erfcinv(2*pfa)-snr(i)*sqrt(n/2))/sqrt(2+4*snr(i))); %awgn channels %pd(i) = marcumq(sqrt(snr(i)*2),sqrt(r),n/2); %pd(i)=pfa+exp(1/((2*n)*(snr(i)^2))-sqrt(2)*erfcinv(2*pfa)/(sqrt(n)*snr(i))) %rayleigh fading(pfa要很小《0.01) end plot(snr1,pd,'o-'); hold on;