Main Content

Recovery of IEEE 802.15.4 OQPSK Signals

This example shows how to implement a practical IEEE® 802.15.4™ PHY receiver decoding OQPSK waveforms that may have been received from wireless radios, using the Communications Toolbox™. This practical receiver has decoded standard-compliant waveforms received from commercial ZigBee radios enabling home automation in the 2.4 GHz band, using a USRP® B200-mini radio and the Communications Toolbox Support Package for USRP® radio add-on.


The IEEE 802.15.4 standard specifies the MAC and PHY layers of low-rate wireless personal area networks (LR-WPANs) [1]. The IEEE 802.15.4 MAC and PHY layers provide the basis of other higher-layer standards, such as ZigBee, WirelessHart®, 6LoWPAN and MiWi. Such standards find application in home automation and sensor networking and are highly relevant to the internet-of-things (IoT) trend.

Receiver Architecture

Overall, the receiver performs the following operations:

  • Matched filtering

  • Coarse frequency compensation

  • Fine frequency compensation

  • Timing Recovery

  • Preamble detection

  • Phase ambiguity resolution

  • Despreading

Between these steps, the signal is visualized to illustrate the signal impairments and the corrections.

Matched Filtering

load lrwpanPHYCaptures % load OQPSK signals from the 2.4 GHz band
spc = 12; % 12 samples per chip
          % the frame was captured at 12x chiprate = 12 MHz

A matched filter improves the SNR of the signal. The 2.4 GHz OQPSK PHY uses half-sine pulses, therefore the following matched filtering operation is needed.

% Matched filter for captured OQPSK signal:
halfSinePulse = sin(0:pi/spc:(spc)*pi/spc);
decimationFactor = 3; % reduce spc to 4, for faster processing
matchedFilter = dsp.FIRDecimator(decimationFactor,halfSinePulse);
filteredOQPSK = matchedFilter(capturedFrame1); % matched filter output

Frequency Offsets

Decoding a signal under the presence of frequency offsets is a challenge for any wireless receiver. Frequency offsets up to 30 kHz were measured for signals transmitted from commercial ZigBee radios and captured using a USRP® B200-mini radio.

Constellation diagrams can illustrate the quality of the received signal, but it is first important to note that the trajectory of an ideal OQPSK signal follows a circle.

% Plot constellation of ideal OQPSK signal
msgLen = 120;                       % length in bytes
message = randi([0 1], msgLen, 1);  % transmitted message
oqpskConfig = lrwpanOQPSKConfig(SamplesPerChip=spc, PSDULength=msgLen);
idealOQPSK = lrwpanWaveformGenerator(message, oqpskConfig);
constellation = comm.ConstellationDiagram( ...
    'Name','Ideal OQPSK Signal', ...
constellation.Position = [constellation.Position(1:2) 300 300];

The above constellation also contains one radius corresponding to the frame start, and one radius corresponding to the frame end. At the same time, frequency offsets circularly rotate constellations, resulting in ring-shaped constellations as well. Therefore, it is more meaningful to observe the constellation of a QPSK-equivalent signal that is obtained by delaying the in-phase component of the OQPSK signal by half a symbol. When half-sine pulse filtering is used, and the oversampling factor is greater than one, the ideal QPSK constellation resembles an 'X'-shaped region connecting the four QPSK symbols (red crosses) with the origin.

% Plot constellation of ideal QPSK-equivalent signal
idealQPSK = helperAlignIQ(idealOQPSK, spc);
constellation.Name = 'Ideal QPSK-Equivalent Signal';
constellation.ReferenceConstellation = [1+1i 1-1i 1i-1 -1i-1];

However, the samples of the captured frame are dislocated from this 'X'-shaped region due to frequency offsets:

% Plot constellation of QPSK-equivalent (impaired)
% received signal
filteredQPSK = helperAlignIQ(filteredOQPSK, spc/decimationFactor);
constellation = comm.ConstellationDiagram( ...
    'XLimits',[-7.5 7.5], ...
    'YLimits',[-7.5 7.5], ...
    'ReferenceConstellation',5*qammod(0:3, 4), ...
    'Name','Received QPSK-Equivalent Signal');
constellation.Position = [constellation.Position(1:2) 300 300];

Coarse Frequency Compensation

Such frequency offsets are first coarsely corrected using an FFT-based method [2] that squares the OQPSK signal and reveals two spectral peaks. The coarse frequency offset is obtained by averaging and halving the frequencies of the two spectral peaks.

% Coarse frequency compensation of OQPSK signal
coarseFrequencyCompensator = comm.CoarseFrequencyCompensator( ...
    'Modulation','OQPSK', ...
    'SampleRate',spc*1e6/decimationFactor, ...
[coarseCompensatedOQPSK, coarseFrequencyOffset] = ...
fprintf('Estimated frequency offset = %.3f kHz\n', ...
Estimated frequency offset = 26.367 kHz
% Plot QPSK-equivalent coarsely compensated signal
coarseCompensatedQPSK = helperAlignIQ(coarseCompensatedOQPSK, spc/decimationFactor);
constellation.Name = ...
    'Coarse frequency compensation (QPSK-Equivalent)';

Some samples still lie outside the 'X'-shaped region connecting the origin with the QPSK symbols (red crosses), as fine frequency compensation is also needed.

Fine Frequency Compensation

Fine frequency compensation follows the OQPSK carrier-recovery algorithm described in [3]. This algorithm is behaviorally different than its QPSK counterpart, which does not apply to OQPSK signals even if their in-phase signal component is delayed by half a symbol.

% Fine frequency compensation of OQPSK signal
fineFrequencyCompensator = comm.CarrierSynchronizer( ...
    'Modulation','OQPSK', ...
fineCompensatedOQPSK = ...

% Plot QPSK-equivalent finely compensated signal
fineCompensatedQPSK = helperAlignIQ(fineCompensatedOQPSK, spc/decimationFactor);
constellation.Name = 'Fine frequency compensation (QPSK-Equivalent)';

The constellation is now closer to its ideal form, but still timing recovery is needed.

Timing Recovery

Symbol synchronization occurs according to the OQPSK timing-recovery algorithm described in [3]. In contrast to carrier recovery, the OQPSK timing recovery algorithm is equivalent to its QPSK counterpart for QPSK-equivalent signals that are obtained by delaying the in-phase component of the OQPSK signal by half a symbol.

% Timing recovery of OQPSK signal, via its QPSK-equivalent version
symbolSynchronizer = comm.SymbolSynchronizer( ...
    'Modulation','OQPSK', ...
syncedQPSK = symbolSynchronizer(fineCompensatedOQPSK);

% Plot QPSK symbols (1 sample per chip)
constellation.Name = 'Timing Recovery (QPSK-Equivalent)';

Note that the output of the symbol synchronizer contains one sample per symbol. At this stage, the constellation truly resembles a QPSK signal. The few symbols that gradually move away from the origin correspond to the frame start and end.

Preamble Detection, Despreading and Phase Ambiguity Resolution:

Once the signal has been synchronized, the next step is preamble detection, which is more successful if the signal has been despreaded. It is worth noting that fine frequency compensation results in a π/2-phase ambiguity, indicating the true constellation may have been rotated by 0, π/2, π, or 3π/2 radians. Preamble detection resolves the phase ambiguity by considering all four possible constellation rotations. The next function operates on the synchronized OQPSK signal, performs joint despreading, resolution of phase ambiguity and preamble detection, and then outputs the MAC protocol data unit (MPDU).

MPDU =  lrwpan.PHYDecoderOQPSKAfterSync(syncedQPSK);
Found preamble of OQPSK PHY.
Found start-of-frame delimiter (SFD) of OQPSK PHY.

Further Exploration

The example uses the lrwpan.PHYDecoderOQPSKAfterSync utility to decode an IEEE 802.15.4 OQPSK waveform. The API and functionality of this undocumented utility may change in the future. To view the source code, use the edit function.


1 - "IEEE Standard for Local and Metropolitan Area Networks--Part 15.4: Low-Rate Wireless Personal Area Networks (LR-WPANs)," in IEEE Std 802.15.4-2011 (Revision of IEEE Std 802.15.4-2006) , vol., no., pp.1-314, 5 Sept. 2011, doi: 10.1109/IEEESTD.2011.6012487.

2 - "Designing an OQPSK demodulator", Jonathan Olds.

3 - Rice, Michael. Digital Communications - A Discrete-Time Approach. 1st ed. New York, NY: Prentice Hall, 2008.

Related Topics