Table of contents
1. Preview of algorithm operation renderings
2.Algorithm running software version
4. Overview of algorithm theory
5. Algorithm complete program engineering
1. Preview of algorithm operation renderings
2.Algorithm running software version
matlab2022a
3. Some core programs
%EM算法收敛所需的迭代
nIter = 1;
Yp = Y(:,1:L_polit,:); %与导频序列相对应的部分
qmse1 = zeros(1,len); %EM算法的MSE
for k=1:len %计算所有SNR的EM算法的MSE
G0 = zeros(Nant,Nuser); %初始状态
mIu0 = zeros(Nuser,N);
sgm0 = zeros(Nuser,Nuser);
for i=1:nIter
G0 = (Yp(:,:,k)*Polits' + Y(:,L_polit:N-1,k)*mIu0(:,L_polit:N-1)')/ (Polits*Polits' + mIu0(:,L_polit:N-1)*mIu0(:,L_polit:N-1)' + (N-L_polit)*sgm0);
for j=1:N
mIu0(:,j) = (G0'*G0 + sigmaNu(k)*eye(Nuser))\(G0'*Y(:,j,k));
end
sgm0 = sigmaNu(k)*eye(Nuser)/(G0'*G0 + sigmaNu(k)*eye(Nuser));
end
qmse1(k) = trace(abs((G-G0)'*(G-G0)))/mean(beta2);
end
nIter = 5;
Yp = Y(:,1:L_polit,:); %与导频序列相对应的部分
qmse2 = zeros(1,len); %EM算法的MSE
for k=1:len %计算所有SNR的EM算法的MSE
G0 = zeros(Nant,Nuser); %初始状态
mIu0 = zeros(Nuser,N);
sgm0 = zeros(Nuser,Nuser);
for i=1:nIter
G0 = (Yp(:,:,k)*Polits' + Y(:,L_polit:N-1,k)*mIu0(:,L_polit:N-1)')/ (Polits*Polits' + mIu0(:,L_polit:N-1)*mIu0(:,L_polit:N-1)' + (N-L_polit)*sgm0);
for j=1:N
mIu0(:,j) = (G0'*G0 + sigmaNu(k)*eye(Nuser))\(G0'*Y(:,j,k));
end
sgm0 = sigmaNu(k)*eye(Nuser)/(G0'*G0 + sigmaNu(k)*eye(Nuser));
end
qmse2(k) = trace(abs((G-G0)'*(G-G0)))/mean(beta2);
end
614. Overview of algorithm theory
The semi-blind channel estimation algorithm based on massive MIMO communication systems involves multiple steps. Its principles and mathematical formulas are summarized as follows:
First, the MIMO system needs to send a known training sequence to perform initial channel estimation at the receiving end. When sending useful information data, the receiving end will use the initial channel estimation results to update the decision to complete real-time channel estimation.
On this basis, the semi-blind channel estimation algorithm combines the characteristics of blind estimation and training sequence-based estimation. Generally speaking, it is a common way to estimate by designing a training sequence or periodically inserting pilot symbols in the data.
The mathematical formula of semi-blind channel estimation can be expressed as:
H^S = (1/T) * Σ_t=1^T [y_t * conj(H_t) / (1 + Σ_i=1^L * conj(H_i) * y_t * conj(H_i)^*)] （5）
Among them, H^S is the estimated result of the massive MIMO channel, y_t is the received signal vector, conj(H_t) is the conjugate transpose of H_t, L is the number of pilot symbols, conj(H_i)^* is the common of H_i The complex conjugate of the yoke transpose.
This formula is based on the idea of blind estimation, using some inherent characteristics of the modulated signal itself that are not related to the specific information bits, or adopting a decision feedback method for channel estimation. At the same time, by inserting known pilot symbols into the sent useful data, the channel estimation result of the pilot position can be obtained, and then the channel estimation result of the pilot position is used to obtain the channel estimation result of the useful data position through interpolation.
Therefore, the semi-blind channel estimation algorithm not only uses the initial estimation based on the training sequence and the real-time decision update, but also combines the characteristics of the blind estimation method to estimate the massive MIMO channel more accurately.
It should be noted that the semi-blind channel estimation algorithm also needs to consider other factors in practical applications, such as training sequence design, selection and insertion of pilot symbols, implementation of decision feedback mechanism, etc. These factors may have an impact on the performance and practical application of the algorithm. Therefore, in practical applications, algorithm optimization and adjustment need to be carried out according to specific conditions.
5. Algorithm complete program engineering
OOOOO
OOO
O