COSPAS-SARSAT T.018 DSSS OQPSK Implementation - Technical Questions

Dear MATLAB Project Team,

We are implementing a C version of your DSSS OQPSK receiver for COSPAS-SARSAT T.018 second-generation beacons ( https://fr.mathworks.com/help/comm/ug/dsss-receiver-for-sar-based-tracking-system.html ). We have successfully replicated most components but are experiencing persistent low correlation values (0.27-0.30) below the detection threshold (0.35) despite

matching all documented parameters.

**1. Signal Generation**

- What exact pulse shaping filter is used in the transmitter?

- Half-sine pulse: `sin(π×n/SPS)` for n=0 to SPS-1?

- Raised cosine with specific rolloff factor?

- No pulse shaping (rectangular)?

- What is the expected signal power/amplitude after modulation (before AGC)?

**2. Receiver Configuration**

- What SPS value was used for hardware testing?

- Documentation shows SPS=8, but PlutoSDR minimum sample rate (2.083 MHz) would require SPS≥55

- Was the MATLAB code tested with real hardware or only in simulation?

- For the `comm.SymbolSynchronizer`:

- Is the default TED "Zero-Crossing" or was a different method specified?

- Are the parameters (NormalizedLoopBandwidth=0.001, DetectorGain=4, DampingFactor=2) valid for both SPS=8 and SPS=16?

**3. Preamble Detection**

- In `helperPolyphaseCorrelator`, the threshold is set to 0.35:

- Is this threshold empirical or calculated from expected SNR?

- What correlation values do you typically observe with clean signals (no noise)?

- Should we expect ~0.8-0.9 correlation for perfect signals?

**4. OQPSK to QPSK Conversion**

- What exact alignment is used when converting OQPSK to QPSK?

- Fixed delay of `sps/2` samples?

- Adjusted based on timing recovery output?

- Phase offset applied (e.g., `sps/2` to capture pulse peaks)?

**5. Matched Filtering**

- Is a matched filter required for half-sine pulse shaped signals?

- If yes, where in the receiver chain (before/after AGC, before/after timing recovery)?

**Our Current Results:**

- Polyphase correlator: max correlation ~0.30, threshold ~0.37 (adaptive: 5.5×mean)

- Signal verification: constant envelope, power=0.5, correct PRN sequences

- All timing/carrier sync parameters match MATLAB documentation

- π/4 rotation applied in both transmitter and preamble reference

Could you provide guidance on what might cause systematically low correlation despite correct implementation of documented parameters?

Thank you for your assistance.

Best regards

[John D'Errico]

Answers is not a MATLAB project team. If you want to contact something/someone like that, you need to contact technical support directly.

https://www.mathworks.com/company/aboutus/contact_us.html?s_tid=gf_contact