QPSK modulation/SER

I have project to finish, which is QPSK modulator/demodulator simulation with matlab. At first I tried using gray coding, but the decided to do using I/Q components, for some reasons. I think modulation part is successful, but when I got to the demodulation part, where I need to find false bits...for my SER. I just can't get desired results. What I tried, is correlating received signals (carry for 1 and carry for 0) and then comparing them both, but this gave me 0 errors... Just look at the part that's correlating signal and tell me if you have any ideas. ty

clc;clear all;close all;
%1 Uzduotis
bit_stream = randi([0 1],[1 10^4]);
Fc = 10^6; 
T_bit = 5*10^(-6);
F_disc = 10^7;
n = 0:(F_disc*T_bit)-1;
t = n/F_disc;
%waveforms for odd bits
I = cos(2*pi*Fc*t);  
%I1 = cos(2*pi*Fc*t+pi);
%waveforms for even bits
Q = sin(2*pi*Fc*t); 
%Q1 = sin(2*pi*Fc*t+pi);
i = 1;
j = 1;
odd_bits = zeros(1,length(bit_stream)/2);
even_bits = zeros(1,length(bit_stream)/2);
while (i<length(bit_stream))
    odd_bits(j) = bit_stream(i);
    even_bits(j) = bit_stream(i+1);
    i = i+2;
    j = j+1;
end
temp_odd = odd_bits;
temp_even = even_bits;

for i = 1:length(odd_bits)
    if odd_bits(i)==0
        odd_bits(i)=-1;
    end
    if even_bits(i)==0
        even_bits(i)=-1;
    end
end
qI = [];
qQ = [];
QPSK_signal = [];
for i = 1:length(odd_bits)
    qI =[qI odd_bits(i)*I];
end
for i = 1:length(even_bits);
    qQ = [qQ even_bits(i)*Q];
end

QPSK_signal = qI+qQ;

figure(1);
subplot(3,1,1);
plot(I);
title('cos wave');
grid on;
subplot(3,1,2);
plot(Q);
title('sin wave');
grid on;
subplot(3,1,3);
stem(bit_stream);
title('Moduliuojamas bitu srautas');
axis([0 200 -0.5 1.5]);

figure(2)
subplot(3,1,1);
plot(qI);
axis([0 200 -1.5 1.5]);
title('Odd bits');
grid on;
subplot(3,1,2);
plot(qQ);
axis([0 200 -1.5 1.5]);
title('Even bits');
grid on;
subplot(3,1,3);
plot(QPSK_signal,'LineWidth',2);
axis([0 200 -1.5 1.5]);
title('QPSK signal');
grid on;

%3 Uzduotis
%figure(3)
%periodogram(QPSK_signal,window,'onesided',10e6,F_disc);

%4 Uzduotis
h_max = 30;
%h = 3;
counth = 0;
j = 1;
s = 1;
rec_sigI = [];
reg_sigI0 = [];
for h = 1:0.1:h_max
    count = 0;
    for i = 1:length(odd_bits)
        Noise = randn(1,length(qI));
        rec_sig = h*qI + Noise;
        %rec_sigI0 = h*(-I)+Noise;
        K1 = sum(rec_sig(1:length(I)).*I);
        K2 = sum(rec_sig(1:length(I)).*(-I));
        if K2 > K1
            count = count+1;
        end
    end
    counth(j) = count;
    j = j+1;
end
ser = counth/10000;
figure();
loglog(ser);
grid on;