Grant: UCSD – Automated Capture and Decoding on Modest Wideband SDRs

Date of Grant: Nov. 6, 2020
Grant Amount: $73,836

Founded in 1961, the University of California, San Diego has rapidly achieved as status of one of the top institutions in the nation for higher education and scientific research. This grant enables one of their PhD Students – Sam Crow, supervised by Dr. Aaron Schulman – to build an open source, wideband (e.g., 20-100 MHz) automated spectrum capture and decoding system. This FPGA-accelerated and software-based receiver will automatically and efficiently capture, decode, and store transmissions from analog and digital sources.

About the Project

The popularity of SDR platforms has led to the creation of a vast array of capture tools and decoders for many of the most common and interesting analog digital waveforms on the spectrum (e.g., AM/FM, ACARS/VDL2, SSTV, etc.). Although these tools to make it possible to find and listen to interesting transmissions on the spectrum, in practice, using them is cumbersome: users still need to manually tune in to each frequency of interest (perhaps guided by a live spectrogram), and select a demodulator/decoder that they want to use on the signal, then they can listen to the activity on that frequency.

The team aims to design a system so that it can operate on modest, widely-available SDR platforms (e.g., Pluto and Lime SDR), so that this capability can be made available to even an entry-level radio hobbyist. The heart of this system will be the resource-efficient, open source, sparsity-based backhaul and compute system for SDRs, called SparSDR.

Developed by the UCSD team, SparSDR significantly reduces the backhaul bandwidth for raw signal captures from wideband SDRs, without sacrificing the ability to capture raw signals. It compresses wideband captures using overlapped, windowed FFT accelerators, and a thresholding scheme, that surprisingly fits into the limited hardware resources of the FPGA in even modest wideband SDRs (e.g., AD Pluto). SparSDR also introduces a novel signal reconstruction algorithm that only requires modest CPU to recover the raw samples from these compressed captures. The UCSD team has already demonstrated the possibility of decoding multi-channel BLE advertisements across 80 MHz, using the low-end CPU on a Raspberry Pi 3. SparSDR’s capabilities have been well received by the radio hobbyist community, indicating that there could be a significant user base if the platform more widely available and easy to use.

The project aims to achieve the following during the course of this project:

  1. Make SparSDR stable and easy to use,
  2. Port SparSDR to other popular SDR platforms,
  3. Create the automated decoding system, and
  4. Make the decoded output easy and intuitive to browse.

For more information about this project, please reach out to Aaron Schulman: