This example shows how to use the COMM.EVMCOMM.EVM System object to measure the error vector magnitude (EVM) of a simulated IEEE® 802.15.4 [ 1 ] transmitter. IEEE 802.15.4 is the basis for the ZigBee specifications.
Error Vector Magnitude (EVM)
The error vector magnitude (EVM) is a measure of the difference between a reference waveform, which is the error-free modulated signal, and the actual transmitted waveform. EVM is used to quantify the modulation accuracy of a transmitter. [ 1 ] requires that a 802.15.4 transmitter shall not have an RMS EVM value worse than 35%.
An 802.15.4 system for 868 MHz band employs direct sequence spread spectrum (DSSS) with binary phase-shift keying (BPSK) used for chip modulation and differential encoding used for data symbol encoding.
dataRate = 20e3; % Bit rate in Hz M = 2; % Modulation order (BPSK) chipValues = [1;1;1;1;0;1;0;1;1;0;0;1;0;0;0]; % Chip values for bit 0. % Chip values for 1 is the opposite.
Section 6.7.3 of [ 1 ] specifies that the measurements are performed over 1000 samples of I and Q baseband outputs. To account for filter delays, we include 1 more bit in the simulation of the transmitted symbols. We chose to oversample the transmitted signal by four. We assume an SNR of 60 dB to account for transmitter and test hardware imperfections.
numSymbols = 1000; % Number of symbols required for one EVM value numFrames = 100; % Number of frames nSamps = 4; % Number of samples that represents a symbol filtSpan = 8; % Filter span in symbols gain = length(chipValues); % Spreading gain (number of chips per symbol) chipRate = gain*dataRate; % Chip rate sampleRate = nSamps*chipRate; % Final sampling rate numBits = ceil((numSymbols)/gain)+1; % Number of bits required for one EVM value SNR = 60; % Simulated signal-to-noise ratio in dB
We can obtain BPSK modulated symbols with a simple mapping of 0 to +1 and 1 to -1. If we also map the chip values, then we can modulate before bit-to-chip conversion and use matrix math to write efficient MATLAB® code. ZigBee specifications also define the pulse shaping filter as having a raised cosine pulse shape with rolloff factor of 1.
% Map chip values chipValues = 1 - 2*chipValues; % Design a raised cosine filter with rolloff factor 1 hTxFilter = comm.RaisedCosineTransmitFilter('RolloffFactor', 1, ... 'OutputSamplesPerSymbol', nSamps, ... 'FilterSpanInSymbols', filtSpan); hRxFilter = comm.RaisedCosineReceiveFilter('RolloffFactor', 1, ... 'InputSamplesPerSymbol', nSamps, ... 'FilterSpanInSymbols', filtSpan, ... 'DecimationFactor', nSamps);
The Communications System Toolbox™ provides COMM.EVM to calculate RMS EVM, Maximum EVM, and Xth percentile EVM values. Section 6.7.3 of [ 1 ] defines the EVM calculation method, where the average error of measured I and Q samples are normalized by the power of a symbol. For a BPSK system, the power of both constellation symbols is the same. Therefore, we can use the 'Peak constellation power' normalization option. Other available normalization options, which can be used with other communications system standards, are average constellation power and average reference signal power.
hEVM = comm.EVM('Normalization', 'Peak constellation power')
hEVM = System: comm.EVM Properties: Normalization: 'Peak constellation power' PeakConstellationPower: 1 MaximumEVMOutputPort: false XPercentileEVMOutputPort: false
We first generate random data bits, differentially encode these bits using a
DifferentialEncoder System object, and modulate using BPSK. We spread the modulated symbols by employing a matrix multiplication with the mapped chip values. The spread symbols are then passed through a pulse shaping filter. The EVM object assumes that received symbols, rd, and reference symbols, c, are synchronized, and sampled at the same rate. We downsample the received signal, r, and synchronize with the reference signal, c.
[ 1 ] requires that 1000 symbols be used in one RMS EVM calculation. To ensure we have enough averaging, we simulate 100 frames of 1000 symbols and use the maximum of these 100 RMS EVM measurements as the measurement result. We see that the simulated transmitter meets the criteria mentioned in Error Vector Magnitude section above.
% Tx and Rx filter delays are identical and equal to half the filter span. % Total delay is equal to the sum of two filter delays, which is the filter % span of one filter. refSigDelay = hTxFilter.FilterSpanInSymbols; % Simulated number of symbols in a frame simNumSymbols = numBits*gain; % Initialize peak RMS EVM peakRMSEVM = -inf; % Create a comm.DifferentialEncoder object to differentially encode data hdiffenc = comm.DifferentialEncoder; % Create an comm.AWGNChannel System object and set its NoiseMethod property % to 'Signal to noise ratio (SNR)' hChan = comm.AWGNChannel('NoiseMethod', 'Signal to noise ratio (SNR)',... 'SNR', SNR); % Loop over bursts for p=1:numFrames % Generate random data b = randi([0 M-1], numBits, 1); % Differentially encode d = step(hdiffenc, b); % Modulate x = 1-2*d; % Convert symbols to chips (spread) c = reshape(chipValues*x', simNumSymbols, 1); % Pulse shape cUp = step(hTxFilter, c); % Calculate and set the 'SignalPower' property of the channel object hChan.SignalPower = sum(cUp.^2)/length(cUp); % Add noise r = step(hChan, cUp); % Downsample received signal. Account for the filter delay. rd = step(hRxFilter, r); % Measure using the EVM System object rmsEVM = step(hEVM, ... complex(rd(refSigDelay+(1:numSymbols))), ... complex(c(1:numSymbols))); % Update peak RMS EVM calculation if (peakRMSEVM < rmsEVM) peakRMSEVM = rmsEVM; end end % Display results fprintf(' Worst case RMS EVM (%%): %1.2f\n', peakRMSEVM)
Worst case RMS EVM (%): 0.19
We showed how to utilize COMM.EVM to test if a ZigBee transmitter complies with the standard specified EVM values. We used a crude model that only introduces additive white Gaussian noise and showed that the measured EVM is less than the standard specified 35%.
IEEE Standard 802.15.4, Wireless Medium Access Control (MAC) and Physical Layer (PHY) Specifications for Low-Rate Wireless Personal Area Networks, 2003.