Awaiting transmission
The receiver acquires the carrier, recovers symbols, and decodes CCSDS frames.
Carrier tracking during the transmission
Frame-by-frame results
| Frame | Sent at s into transmission | Predicted Es/N0 dB | Channel bit errors | Bit errors after Viterbi | Reed–Solomon | Result |
|---|
The SigMF recording holds the first 262,144 complex samples after Doppler pre-correction, at four samples per symbol. GNU Radio, inspectrum and SigMF tools can open it.
—
Link budget at the start of the transmission.
| Term | Value | Notes |
|---|
Link margin over the selected pass
Doppler shift over the selected pass
Sky track azimuth and elevation from the station
| # | Rise (UTC) | Max elevation | Duration | Positive margin | Peak margin dB | Data volume |
|---|
Geometry
SGP4 propagates the selected TLE, OMM or reference elements in the TEME frame. The ground station sits on the WGS-84 ellipsoid and rotates with Greenwich mean sidereal time; polar motion is ignored. Range rate comes from the relative velocity, including Earth rotation at the station. Passes are found on a coarse time grid, with rise and set refined by bisection and maximum elevation by golden-section search. Body-fixed spacecraft antennas are assumed to face nadir.
Link budget
Free-space loss uses the slant range. Clear-air absorption is a zenith value of about 0.04–0.07 dB between UHF and X band, scaled by the Kasten–Young air mass. Rain attenuation follows the simplified ITU-R P.618 slant-path method with ITU-R P.838-3 coefficients for circular polarization. The rain height falls with latitude after ITU-R P.839. Antenna noise adds a cold-sky galactic background scaled from 408 MHz, sky emission from the absorbing atmosphere and rain at 275 K, ITU-R P.372 median man-made noise for the chosen site type, and a fixed ground spillover (10 K for a dish, 40 K for a Yagi). System temperature is referred to the LNA input. Spacecraft antennas use a cosine-power patch pattern, an isoflux pattern shaped to equalize slant range to 5° elevation, a constant-gain omnidirectional pattern, or a steered beam with a Gaussian pointing loss. Ground antennas use the same pointing-loss rule with a beamwidth of 70 λ/D for a dish.
Eb is the energy per transfer-frame bit, so sync-marker and coding overhead count against the link. The required Eb/N0 is the level at which this tool's own receiver loses 1% of frames on a static channel, measured by scripts/space/calibrate_rf_thresholds.cjs. Margin subtracts that requirement and the implementation loss you enter.
Transmitter and channel
The payload is a 128 × 72 color image split into CCSDS-style transfer frames of 1,115 bytes, each with a frame counter and a CRC-16 frame error control field. Frames are Reed–Solomon (255, 223) encoded at interleaving depth 5 with the CCSDS field and generator (conventional basis, without the dual-basis transform), randomized with the CCSDS pseudo-random sequence, prefixed with the 0x1ACFFC1D attached sync marker, and convolutionally encoded with the CCSDS (7, ½) code. One idle frame leads and one trails the image. PSK symbols use root-raised-cosine pulses with roll-off 0.35; GMSK uses BT 0.5 and modulation index 0.5.
The channel is generated at four samples per symbol. Every sample is delayed by the one-way propagation time R(t)/c, interpolated from SGP4 ranges and range rates. Carrier Doppler, Doppler rate and the matching stretch of the symbol clock all come from that delay. The receiver's own orbit prediction, optionally offset along track, is subtracted as Doppler pre-correction. The spacecraft oscillator adds a fixed frequency offset and Wiener phase noise whose level at 1 kHz offset you set. Gaussian noise follows the link budget's Es/N0, updated along the transmission as range, elevation and antenna gain change. Signals that the pre-correction leaves outside the receiver's sample bandwidth are not received.
Receiver
PSK: the matched-filter output is hard-limited and raised to the M-th power, which turns the modulated carrier into a spectral line. Up to 24 FFTs of 8,192 points are averaged, and only the acquisition range you set is searched. The symbol clock leaves weaker lines at ±Rs/M on either side of the carrier, so the receiver picks the line that has both neighbors. The detection threshold allows one false acquisition in about 1,000 attempts. Feed-forward estimates from the first 1,024 symbols set the starting sample phase and carrier phase. A Gardner detector with cubic interpolation then tracks symbol timing, starting four times wider and narrowing after 1,500 symbols. A second-order decision-directed loop tracks the carrier at the bandwidth you set; its frequency integrator waits 400 symbols for timing to settle, which avoids false locks at low SNR. Moment-based (M2M4) estimates give Es/N0 and scale the soft decisions. GMSK: the receiver finds the signal's spectral centroid, then uses a non-coherent limiter–discriminator with one-symbol integration and Gardner timing. Soft decisions are clipped to limit the effect of discriminator clicks. This detector costs several decibels against coherent detection, and the convolutional code helps it less than Reed–Solomon does.
The decoder tries every carrier-phase rotation and symbol-pair alignment, keeps the one that finds the sync marker, then runs soft-decision Viterbi decoding, frame synchronization, derandomization, Reed–Solomon decoding and the CRC check. Like a hardware frame synchronizer, it reads through one or two missed sync markers (a flywheel) and lets Reed–Solomon and the CRC judge those frames. After three misses in a row it counts a re-acquisition and searches again. Frames that pass the CRC are placed by their frame counter. The simulation knows what was sent, so it reports true bit errors before and after decoding, compared with theory for an ideal coherent receiver at the predicted Es/N0. The GMSK comparison is the coherent MSK bound.
Model limits
This is a screening and teaching tool. Atmospheric and rain models are simplified, and the rain rate is applied as a steady value rather than an exceedance statistic. Ionospheric scintillation, Faraday rotation, multipath, interference, amplifier nonlinearity and adaptive coding are not modeled. Antenna patterns are idealized and ignore spacecraft attitude other than nadir pointing. Each run simulates one image transmission starting at the selected time, not a continuous pass. Thresholds come from 200-frame simulations in 0.25 dB steps and carry about ±0.5 dB of statistical uncertainty; in theory BPSK and QPSK are identical. Coded 8PSK is limited by blind carrier acquisition and tracking near 7.5 dB Es/N0 rather than by the code, so adding Reed–Solomon gains it little. Real high-order PSK modems insert pilot symbols to avoid this.
CCSDS Blue Books (131.0-B, TM synchronization and channel coding) · CCSDS Green Books (130.1-G, coding rationale) · ITU-R P.618 · ITU-R P.838 · ITU-R P.372 · F. M. Gardner, “A BPSK/QPSK timing-error detector for sampled receivers,” IEEE Trans. Communications, 1986 · SigMF · SGP4 frames