这个示例展示了如何使用
通信系统工具箱,通过使用眼图和散点图来可视化信号行为。该示例使用QPSK信号,该信号通过方根升余弦(RRC)过滤器传递。
Scatter Plot
Set the RRC filter, modulation scheme, and plotting parameters.
span = 10; % Filter span
rolloff = 0.2; % Rolloff factor
sps = 8; % Samples per symbol
M = 4; % Modulation alphabet size
k = log2(M); % Bits/symbol
phOffset = pi/4; % Phase offset (radians)
n = 1; % Plot every nth value of the signal
offset = 0; % Plot every nth value of the signal, starting from offset+1
Create the filter coefficients using the rcosdesign function.
filtCoeff = rcosdesign(rolloff,span,sps);
Generate random symbols for an alphabet size of M.
rng default
data = randi([0 M-1],5000,1);
Apply QPSK modulation.
dataMod = pskmod(data,M,phOffset);
Filter the modulated data.
txSig = upfirdn(dataMod,filtCoeff,sps);
Calculate the SNR for an oversampled QPSK signal.
EbNo = 20;
snr = EbNo + 10*log10(k) - 10*log10(sps);
Add AWGN to the transmitted signal.
rxSig = awgn(txSig,snr,'measured');
Apply the RRC receive filter.
rxSigFilt = upfirdn(rxSig, filtCoeff,1,sps);
Demodulate the filtered signal.
dataOut = pskdemod(rxSigFilt,M,phOffset,'gray');
Use the scatterplot function to show scatter plots of the signal before and after filtering. You can see that the receive filter improves performance as the constellation more closely matches the ideal values. The first span symbols and the last span symbols represent the cumulative delay of the two filtering operations and are removed from the two filtered signals before generating the scatter plots.
h = scatterplot(sqrt(sps)*txSig(sps*span+1:end-sps*span),sps,offset,'g.');
hold on
scatterplot(rxSigFilt(span+1:end-span),n,offset,'kx',h)
scatterplot(dataMod,n,offset,'r*',h)
legend('Transmit Signal','Received Signal','Ideal','location','best')
Eye Diagram
Display 1000 points of the transmitted signal eye diagram over two symbol periods.
eyediagram(txSig(sps*span+1:sps*span+1000),2*sps)
Display 1000 points of the received signal eye diagram.
eyediagram(rxSig(sps*span+1:sps*span+1000),2*sps)
Observe that the received eye diagram begins to close due to the presence of AWGN. Moreover, the filter has finite length which also contributes to the non-ideal behavior.