Open-Source Satellite Communication & Signal Processing Toolkit
DynamiX Labs started out as a weekend experiment — just trying to pull down a NOAA weather satellite image with a cheap RTL-SDR dongle and a homemade dipole antenna taped to a window. It worked (barely), and that one blurry APT image kicked off everything you see here.
The idea was simple: build an end-to-end ground station pipeline that could go from raw RF to decoded telemetry, without needing a $50k setup or a PhD to get it running. Something a college student could plug in and start receiving real satellite data on day one.
Over time it grew into four separate tools that all talk to each other — an SDR channelizer, a Doppler tracker, a telemetry decoder, and a hardware benchmark suite. This repo ties them all together under one roof.
Who is this for? Students setting up their first ground station, amateur radio operators messing with satellite comms, CubeSat teams who need a decoder they can actually modify, or anyone who thinks pulling data out of thin air is cool (it is).
Here's the big picture. The signal comes off the antenna, gets digitized by whatever SDR hardware you have, passes through the DSP pipeline, and comes out the other end as clean decoded packets. Each layer is its own module so you can swap things in and out without breaking the rest.
flowchart TB
%% Core Styling Directives
classDef hardware fill:#0f172a,stroke:#3b82f6,stroke-width:2px,color:#e2e8f0,stroke-dasharray: 5 5
classDef tracking fill:#064e3b,stroke:#10b981,stroke-width:2px,color:#e2e8f0
classDef dsp fill:#1e3a8a,stroke:#60a5fa,stroke-width:2px,color:#e2e8f0
classDef telemetry fill:#312e81,stroke:#a78bfa,stroke-width:2px,color:#e2e8f0
classDef ai fill:#4a044e,stroke:#f472b6,stroke-width:2px,color:#e2e8f0
classDef database fill:#451a03,stroke:#fbbf24,stroke-width:2px,color:#e2e8f0
classDef external fill:#171717,stroke:#a3a3a3,stroke-width:2px,color:#d4d4d4
classDef gpu fill:#1a1a2e,stroke:#e94560,stroke-width:2px,color:#eee
classDef ws fill:#16213e,stroke:#0f3460,stroke-width:2px,color:#eee
classDef zmq fill:#0a3d62,stroke:#38ada9,stroke-width:2px,color:#eee
%% External Systems
subgraph External_Network [Global & Space Interfaces]
direction LR
SAT(("fa:fa-satellite Low Earth Orbit\nSatellites")):::external
TLE[("fa:fa-cloud CelesTrak / Space-Track\nREST API (HTTPS)")]:::database
FED_NET(("fa:fa-network-wired DynamiX Federation\nDecentralized Nodes")):::external
end
%% RF Frontend
subgraph RF_Layer [L0: RF Frontend & Digitization]
direction TB
ANT("fa:fa-satellite-dish Az/El Yagi Array\n(VHF/UHF/L-Band)"):::hardware
LNA("fa:fa-bolt Low Noise Amplifier\n(NF < 0.5dB)"):::hardware
HW["fa:fa-microchip SDR Digitizer\n(RTL-SDR / HackRF / USRP)"]:::hardware
BM("fa:fa-stopwatch Hardware Benchmark\nZero-Copy Memory Access"):::hardware
COH("fa:fa-layer-group Coherent Combiner\nRing Buffer (ZMQ IPC)"):::hardware
ANT -- RF Analog --> LNA
LNA -- Amplified RF --> HW
HW == "Complex64 IQ (USB 3.0)" ==> BM
BM == "Stream Filter" ==> COH
end
%% Tracking Engine
subgraph Auto_Tracking [L1: Autonomous Pass Engine]
direction TB
DT["fa:fa-compass Doppler-Auto-Tracker\nSkyfield / SGP4 Predictor"]:::tracking
PID["fa:fa-cogs PID Rotator Controller\n(Hamlib Protocol)"]:::tracking
EMA["fa:fa-wave-square EMA Doppler Filter\nContinuous Tuning"]:::tracking
DT -->|Target Vector| PID
DT -->|Shift Hz| EMA
end
%% GPU-Accelerated DSP Pipeline
subgraph DSP_Layer [L2: GPU-Accelerated DSP]
direction TB
GPU["fa:fa-bolt GPU Backend\nCuPy/CUDA FFT Offload"]:::gpu
SU["fa:fa-filter SatSDR-Universal\nMulti-Band Channelizer"]:::dsp
SPEC["fa:fa-chart-bar Spectral Engine\nGPU Welch PSD & Auto-Detect"]:::dsp
SYNC["fa:fa-sync Carrier/Symbol Sync\nCostas Loop & Gardner TED"]:::dsp
FEC["fa:fa-random FEC Decoder\nViterbi / Reed-Solomon"]:::dsp
COH == "Multi-Band IQ (20 MSPS)" ==> SU
GPU -.->|"Offloaded FFT/FIR"| SU
GPU -.->|"Offloaded PSD"| SPEC
SU == "Isolated Baseband" ==> SPEC
SPEC == "Modulation Class" ==> SYNC
SYNC == "Soft Symbols" ==> FEC
end
%% Telemetry & Security
subgraph Telemetry_Layer [L3: Telemetry, AI, & Security]
direction TB
CTD["fa:fa-shield-alt CubeSat-Telemetry-Decoder\nDeframer (Sync Word)"]:::telemetry
CRYPTO["fa:fa-key Cryptography Engine\nXTEA Decryption & CSP"]:::telemetry
ANOMALY["fa:fa-brain AI Anomaly Detection\nIsolation Forest (TensorFlow)"]:::ai
PKI["fa:fa-lock ECDSA PKI Signer\nSECP256R1 Private Key"]:::telemetry
FEC == "Raw Bitstream" ==> CTD
CTD == "KISS / CSP Frames" ==> CRYPTO
CRYPTO == "Parsed Telemetry" ==> ANOMALY
CRYPTO == "Verified Payload" ==> PKI
end
%% WebSocket Streaming
subgraph Streaming_Layer [L4: WebSocket Live Spectrum Streaming]
direction LR
WSSRV["fa:fa-broadcast-tower Spectrum Server\nasyncio + websockets (port 8765)"]:::ws
PROTO["fa:fa-file-code Stream Protocol\nMessagePack Binary Frames"]:::ws
BROWSER["fa:fa-desktop Browser Dashboard"]:::ws
REMOTE["fa:fa-globe Remote Monitor"]:::ws
SPEC -.->|"PSD + Detections"| WSSRV
WSSRV --> PROTO
PROTO --> BROWSER
PROTO --> REMOTE
end
%% Distributed Decoder Cluster
subgraph Cluster_Layer [L5: ZeroMQ Distributed Decoder Cluster]
direction LR
BROKER["fa:fa-server Decoder Broker\nROUTER/DEALER (5555/5556)"]:::zmq
W1["fa:fa-cog Worker 1\nAPT + ADS-B"]:::zmq
W2["fa:fa-cog Worker 2\nAX.25 + LRPT"]:::zmq
WN["fa:fa-cog Worker N\nGPU-Enabled"]:::zmq
FEC -.->|"Decode Jobs"| BROKER
BROKER -->|"Dispatch"| W1
BROKER -->|"Dispatch"| W2
BROKER -->|"Dispatch"| WN
W1 & W2 & WN -.->|"Results"| BROKER
BROKER -.->|"Aggregated"| WSSRV
end
%% Cross-Layer Integrations
SAT -. "137MHz - 2.4GHz" .-> ANT
TLE -. "Daily Sync" .-> DT
EMA == "Freq Offset" ==> HW
PID == "Az/El Serial" ==> ANT
PKI == "Signed JSON / ZMQ" ==> FED_NET
Each subsystem lives in its own directory and works independently, but they're designed to plug into each other when you need the full pipeline.
| Project | What it does | Status |
|---|---|---|
| SatSDR-Universal | The main SDR engine. Takes raw IQ samples and figures out what signal you're looking at — NOAA weather, ADS-B aircraft, CubeSat beacons, whatever. It handles channelization, spectral detection, and can juggle multiple SDR dongles at once if you've got them. | Active |
| CubeSat-Telemetry-Decoder | Takes demodulated bits and turns them into actual telemetry packets. Handles AX.25, CCSDS, and CSP framing. Also does XTEA decryption if the satellite uses it, and has an anomaly detector that flags weird readings so you're not staring at logs all day. | Active |
| Doppler-Auto-Tracker | Satellites move fast — a LEO pass might shift your signal by ±5 kHz. This tool grabs fresh TLE data, predicts where the bird will be, and continuously nudges your SDR frequency to stay locked on. Also drives antenna rotators via Hamlib if you've got a motorized setup. | Active |
| SDR-Hardware-Benchmark | Before you commit to a particular SDR dongle, you probably want to know how it actually performs. This runs throughput tests, measures dropped samples, checks CPU load, and compares SNR across different devices so you can pick the right one for your use case. | Active |
The benchmark tool spits out JSON and CSV reports — throughput, CPU load, SNR, EVM, all of it. Check SDR-Hardware-Benchmark/README.md for example runs.
We've tested this against live satellite passes. Not simulations, not recorded IQ files (well, those too for regression), but actual signals coming off the antenna. Here's what we got.
SDR Console showing a live wideband spectrum view with the satellite tracking panel locked onto a NOAA-19 pass. The waterfall shows multiple signal carriers across the band with the AOS countdown active.
Gqrx displaying a wideband waterfall scan across the 40m HF band. Multiple signal bursts are visible above the noise floor, demonstrating the spectral isolation capability before demodulation.
Live CAS-4B satellite housekeeping telemetry decoded in real time — RF forward power at 153 mW, OBC temperature 13°C, primary bus voltage 11.03V, 207 mA current draw. 193 packets captured across a single pass, showing the frame decoder extracting clean structured fields from raw beacon frames.
Wideband waterfall showing multiple narrowband carriers above the noise floor. The spectral engine identifies each peak for Doppler correction and frequency locking during satellite passes
Everything talks to the SDR through SoapySDR, so in theory any supported device should work. In practice, here's what we've actually tried and can vouch for:
- RTL-SDR v3 / v4 — cheapest option, great for getting started. Handles NOAA APT and ADS-B just fine.
- HackRF One — wider bandwidth, can transmit too. Good for scanning large chunks of spectrum.
- ADALM-PLUTO — does L-band and has Tx/Rx, handy for Inmarsat experiments.
- USRP B200 / B210 — this is where it gets serious. Full duplex, MIMO, enough bandwidth for HRPT.
- USRP X310 — research-grade. Overkill for most people, but if you've got access to one, it's beautiful.
- LimeSDR Mini — decent middle ground, handles multiple protocols well.
If your SDR isn't on this list but works with SoapySDR, give it a try — it'll probably work.
To quickly configure drivers and libraries for your SDR hardware, run our automated setup script or follow the step-by-step instructions in the hardware-setup/ folder.
- Supports RTL-SDR, HackRF, ADALM-Pluto, USRP (UHD), and LimeSDR.
- Configures udev rules for non-root USB access and establishes the SoapySDR software interfaces.
Here's what the pipeline can currently decode (basic pipelines are working, some of the more exotic modes are still being built out):
- Weather satellites — NOAA APT at 137 MHz, METEOR LRPT at 137.1 MHz, and NOAA HRPT for the higher-res stuff
- Aviation — ADS-B on 1090 MHz (aircraft position/altitude), ACARS on 129.125 MHz (text messages from planes)
- Spacecraft — CubeSat beacons using AX.25, CCSDS, or CSP framing
- Navigation & comms — GPS L1 C/A (just acquisition for now), Inmarsat AERO/STD-C, and Iridium burst detection
We're actively adding more. If there's a protocol you'd like to see supported, open an issue — or better yet, a PR.
Don't want to install GNU Radio and a dozen system packages by hand? Fair enough. There's a Dockerfile in the docker/ directory that sets up everything — Ubuntu 24.04, GNU Radio, SoapySDR, Hamlib, the whole stack.
cd docker
docker build -t dynamix-labs .
docker run -it --device=/dev/bus/usb dynamix-labsYou'll need to pass through your USB device so the container can talk to your SDR dongle. Note that udev rules must be installed on the HOST machine, not just inside the container. If you get permission errors, ensure your host system has the correct rules installed (see the Quick Hardware Setup above) and consider using privileged: true or device_cgroup_rules in your docker-compose.
- @ARYA-mgc — System and DAP Design Engineer
- @Nithi-tech — Software Developer
- @ashwinr-act-cit — RF Engineer
- @jayarajMd — Hardware Engineer
We'd love more people involved. Check out CONTRIBUTING.md if you want to help, and SECURITY.md if you find something that looks like a vulnerability (please report it privately).