Lecture 5
Lecture 5
Lecture 5
- No tags were found...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
MIMO Channel Equalisation MATLAB Programming Conclusions<br />
DSP Applications for Wireless Communications:<br />
Linear Equalisation of MIMO Channels<br />
Dr. Waleed Al-Hanafy<br />
waleed alhanafy@yahoo.com<br />
Faculty of Electronic Engineering, Menoufia Univ., Egypt<br />
Digital Signal Processing (ECE407) — <strong>Lecture</strong> no. 5<br />
August 10, 2011<br />
Dr. Waleed Al-Hanafy Digital Signal Processing (ECE407) — <strong>Lecture</strong> no. 5<br />
Linear Equalisation of MIMO Channels
MIMO Channel Equalisation MATLAB Programming Conclusions<br />
Overview<br />
1 MIMO Channel Equalisation<br />
2 MATLAB Programming<br />
3 Conclusions<br />
Dr. Waleed Al-Hanafy Digital Signal Processing (ECE407) — <strong>Lecture</strong> no. 5<br />
Linear Equalisation of MIMO Channels
MIMO Channel Equalisation MATLAB Programming Conclusions<br />
System Model<br />
v<br />
s<br />
y ˜s<br />
H W H<br />
y = Hs+v (1)<br />
y — noisy N r-dimensional received vector<br />
s — N t-dimensional transmit vector ∈ S N is assumed to be a<br />
spatially-uncorrelated and uniformly distributed complex random vector process<br />
with zero-mean and variance σ 2 s (i.e. R ss = E [ ss H] = σ 2 sI Nt<br />
v — noise vector with dimension N r ×1 drawn from CN ( 0,σ 2 v)<br />
, or<br />
equivalently R vv = E [ vv H] = σ 2 vI Nr<br />
H — N r ×N t flat-fading channel with entries h ij are assumed i.i.d. complex<br />
Gaussian random variables with zero-mean and unit-variance E [ |h ij| 2] = 1, i.e.,<br />
h ij ∈ CN (0,1)<br />
Dr. Waleed Al-Hanafy Digital Signal Processing (ECE407) — <strong>Lecture</strong> no. 5<br />
Linear Equalisation of MIMO Channels
MIMO Channel Equalisation MATLAB Programming Conclusions<br />
Zero-Forcing (ZF) Equaliser<br />
The ZF approach is a very simple linear filter scheme that is accomplished by<br />
computing the Moore-Penrose pseudo-inverse of the channel matrix, H + . Accordingly,<br />
nulling in the ZF criterion is equivalent to completely cancelling the interference<br />
contributed by streams of other users (transmitting antennas). However, ZF receivers<br />
generally suffer from noise enhancement. The ZF filter is therefore given by<br />
( −1H<br />
WZF H = H+ = H H) H H , (2)<br />
and its output as<br />
˜s ZF = W H ZF y = s+ (H H H) −1H H v (3)<br />
The error covariance matrix is therefore given by<br />
Φ ZF = E<br />
[<br />
(˜s ZF −s)(˜s ZF −s) H] ( −1<br />
= σv<br />
2 H H) H (4)<br />
Noticeably, by referring to (4), it is evident that small eigenvalues of H H H will lead to<br />
significant errors due to noise amplification. This, in fact, represents the main<br />
drawback of the ZF filter design as it disregards the noise term from the overall design<br />
and focuses only on perfectly removing the interference term from signal s<br />
Dr. Waleed Al-Hanafy Digital Signal Processing (ECE407) — <strong>Lecture</strong> no. 5<br />
Linear Equalisation of MIMO Channels
MIMO Channel Equalisation MATLAB Programming Conclusions<br />
Minimum Mean Square Error (MMSE) Equaliser<br />
An improved performance over ZF can be achieved by considering the noise term in<br />
the design of the linear filter W H . This is achieved by the MMSE equaliser, whereby<br />
the filter design accounts for a trade-off between noise amplification and interference<br />
suppression. The MMSE filter is obtained by solving for error minimisation of the error<br />
criterion defined by [ ] ( [<br />
ϕ = E e H e = tr E ee H]) (5)<br />
where the error vector e d = s−˜s = s−W H y. Minimisation of ϕ leads to the<br />
Wiener-Hopf equation<br />
WMMSE H Ryy = Rsy (6)<br />
This equation can also be obtained directly by invoking the orthogonality principle<br />
which states that the estimate ˜s achieves minimum mean square error if the error<br />
sequence e is orthogonal to the observation y, i.e., their cross-correlation matrix has to<br />
be the zero matrix E [ ey H] = 0. After some algebraic manipulation, the linear<br />
MMSE-sense filter is given by<br />
( ) −1<br />
WMMSE H = H H H+ σ2 v<br />
σs<br />
2 I Nt H H (7)<br />
Dr. Waleed Al-Hanafy Digital Signal Processing (ECE407) — <strong>Lecture</strong> no. 5<br />
Linear Equalisation of MIMO Channels
MIMO Channel Equalisation MATLAB Programming Conclusions<br />
MMSE Equaliser (cont’d)<br />
Similar to (4) the error covariance matrix using the MMSE filter in (7) is given as<br />
( ) −1<br />
Φ MMSE = σv<br />
2 H H H+ σ2 v<br />
σs<br />
2 I Nt (8)<br />
Obviously by comparing (8) and (4), the error rate of the MMSE solution Φ MMSE is<br />
less than its ZF counterpart Φ ZF specifically at low SNR defined as<br />
[ ]<br />
E ‖s‖ 2 2<br />
SNR = [ ] = tr(Rss)<br />
E ‖v‖ 2 tr(R = σ2 sN t<br />
2 vv) σvN 2 = P budget/N 0 (9)<br />
r<br />
where P budget is the total transmit power budget and N 0 is the total noise power at<br />
the receiver. At high SNR, the second term in (8) will vanish, which leads to<br />
asymptotic error performance similar to the ZF filter. Compared to the ZF filter in (2),<br />
the MMSE filter in (7) can be viewed as a“regularised”expression by a diagonal matrix<br />
of entries σ2 v<br />
σs<br />
2 , which is equal to the reciprocal of the SNR in (9) for equal numbers of<br />
transmit and receive antennas. This regularisation introduces a bias that gives a much<br />
more reliable result than (2) when the matrix is ill-conditioned, A matrix is said to be<br />
ill-conditioned if its condition number (the absolute ratio between the maximum and<br />
minimum eigenvalues) is too large., and/or the estimation of the channel is noisy.<br />
Dr. Waleed Al-Hanafy Digital Signal Processing (ECE407) — <strong>Lecture</strong> no. 5<br />
Linear Equalisation of MIMO Channels
MIMO Channel Equalisation MATLAB Programming Conclusions<br />
MATLAB Programing<br />
clc; clear all;<br />
nt = 4; nr = 4; d = 1000; MinErrs = 100; MaxData = 1e6; chs = 10;<br />
snr dB = 0:3:21; snr = 10.ˆ(snr dB/10); Lsnr = length(snr);<br />
berZF = zeros(1,Lsnr); berMMSE = zeros(1,Lsnr);<br />
for e = 1:chs<br />
disp([’ch ’ num2str(e) ’/’ num2str(chs)]);<br />
H = (randn(nr,nt)+sqrt(-1)*randn(nr,nt))/sqrt(2);<br />
G ZF = inv(H’*H)*H’;<br />
nErZF = zeros(1,Lsnr); nErMMSE = zeros(1,Lsnr); data = zeros(1,Lsnr);<br />
for z = 1:Lsnr<br />
disp([’ step ’ num2str(z) ’/’ num2str(Lsnr) ’: SNR(dB) = ’ num2str(snr dB(z))]);<br />
G MMSE = inv(H’*H + (1/snr(z))*eye(nt))*H’;<br />
while((nErZF(z)
MIMO Channel Equalisation MATLAB Programming Conclusions<br />
MATLAB Programing (cont’d)<br />
aTempZF = G ZF*r; aTempMMSE = G MMSE*r;<br />
aHatZF = FnDecodeQAM(aTempZF,’QPSK’,1);<br />
aHatMMSE = FnDecodeQAM(aTempMMSE,’QPSK’,1);<br />
[dumy dumy nBerZF dumy] = FnSerBerQAM(a,aHatZF,4,’gray’);<br />
[dumy dumy nBerMMSE dumy] = FnSerBerQAM(a,aHatMMSE,4,’gray’);<br />
nErZF(z) = nErZF(z) + nBerZF;<br />
nErMMSE(z) = nErMMSE(z) + nBerMMSE;<br />
end<br />
berZF(z) = berZF(z) + nErZF(z)/data(z);<br />
berMMSE(z) = berMMSE(z) + nErMMSE(z)/data(z);<br />
end<br />
end<br />
berZF = berZF/chs; berMMSE = berMMSE/chs;<br />
semilogy(snr dB,berZF,’k-s’,snr dB,berMMSE,’m-o’); grid on<br />
legend(’ZF’,’MMSE’); grid on; xlabel(’SNR (dB)’); ylabel(’BER’);<br />
.<br />
Dr. Waleed Al-Hanafy Digital Signal Processing (ECE407) — <strong>Lecture</strong> no. 5<br />
Linear Equalisation of MIMO Channels
MIMO Channel Equalisation MATLAB Programming Conclusions<br />
Results<br />
A 4×4 MIMO system with QPSK transmission is considered. It is clearly noted that<br />
the MMSE filter outperforms its ZF counterpart and their BER performance converges<br />
at higher SNR where the regularisation factor σ 2 v/σ 2 s ≪ 1<br />
Nt = 4, Nr = 4, NumChan = 300<br />
10 0 SNR [dB]<br />
ZF<br />
MMSE<br />
10 −1<br />
10 −2<br />
BER<br />
10 −3<br />
10 −4<br />
0 5 10 15 20 25 30<br />
Dr. Waleed Al-Hanafy Digital Signal Processing (ECE407) — <strong>Lecture</strong> no. 5<br />
Linear Equalisation of MIMO Channels
MIMO Channel Equalisation MATLAB Programming Conclusions<br />
Conclusions<br />
Concluding remarks<br />
Linear equalisation systems for linear MIMO systems is studied<br />
Both ZF and MMSE equalisers are derived and analysed<br />
Simulation results comparisons are shown<br />
.<br />
Dr. Waleed Al-Hanafy Digital Signal Processing (ECE407) — <strong>Lecture</strong> no. 5<br />
Linear Equalisation of MIMO Channels