Passive Radar's Grant Application to the Astera Foundation

Describe what you intend to create during your residency in 2 sentences or less (300-character limit)

I’ve developed passive radar code and a $500 hardware kit with a +30 mile range. The residency will allow me to deploy and network ~1000 units, providing open-source sensing infrastructure covering ~10% of the contiguous United States as a public good, giving data for research and model training.

Technically speaking, how will you achieve this? What steps, tools, and resources are required?

Software is in beta, hardware is production-ready, with vendors and logistics in place. We’ve already deployed six radar nodes, with a central server aggregating data.

Steps:
Distribute hardware to users: More nodes to expand the detection area, increase precision, and improve data redundancy/robustness.

Scale server: Analytical complexity grows non-linearly with node count. We'll simulate the full-sized network ahead of full deployment to prepare for scale.

Continuously improve software: As radars are deployed to new users and contexts, learnings will be integrated into data and user workflows.

Validate: Radars detect known aircraft as calibration targets, and as the network grows, we'll confirm the data is improving. We'll test sensing protocols beyond aircraft, such as wildlife, the ionosphere, meteors, and drones.

Resources to add in order to 200x in 12-18 months:
-Subsidize some or all node cost: (~$500) and installation (~$500)
-Add Staff: 1 dev, 1 operations, 3-5 installers/user support

Why are you the right person/team to work on this project? How much progress have you made?

I'm a hardware engineer, previously in nuclear fusion, successful in past entrepreneurship, and open-source projects.

Since starting work on passive radar 18 months ago, I’ve built from scratch:
-Validated, mass producible radar hardware for <$400
-Turnkey, open-source radar OS, first ever on a single board computer
-A server to aggregate and deliver data
Organizational groundwork laid:
-Grants from the ARDC foundation and Mercatus Center, rather than investments to maintain public good alignment
-Legal strategy by Thomsen and Burke LLP
-Partnering with the University of British Columbia, 4 engineering interns contributing
-A team of one full-time developer, plus two part-time specialists (embedded systems and cloud infrastructure)
-Validated radar and server with 6 nodes around North America

We've gone from concept to a live radar network in less than 18 months and $80k. It's the only open-source project of its kind in the space; no one else will build a nationwide network as a public good.

Why is this problem important?

The atmosphere above us is a closed dataset, monitored by militaries but kept classified because the data would reveal their capabilities.

New applications for passive radar are continually being discovered by scientists, who struggle to develop and deploy suitable equipment themselves. With my public passive radars, each researcher wouldn’t need to build one from scratch.

We know passive radar can be used for atmospheric monitoring, wildlife tracking, meteor detection, and geological mapping. There’s now growing interest in tracking drones in civilian airspace, as drone costs drop and capabilities increase.

Beyond monitoring, the network generates massive data sets for model training. The first successful AIs used self-generated data, while text data from the internet gave rise to LLMs. The next class of models will use physical data from real-world measurements.

What the internet did for human communication and knowledge, universal distributed sensing can do for planetary understanding.

Who do you envision using your residency’s outputs, and how?

Researchers: By avoiding the slow and expensive hardware iteration process, they would gain access to data instantly. Space, atmospheric, zoological, and geological sciences would benefit first.

Aviation: Flight data streams are already a demonstrated use case, e.g., ADS-B transponder-based flight-tracking networks. However, radar goes one step further by detecting anything with a radar cross-section, since planes don't always broadcast their location.

UAV Monitoring: Civilian operators can track agricultural and surveillance drones, as well as airborne scientific instrumentation.

AI Builders: Open datasets can train neural nets, and physics-informed neural nets (PINNs) are particularly well-suited here. Users can deploy kits themselves to collect out-of-sample data or target specific use cases. It’s like we’re building a LIGO for radio waves instead of gravitational waves. Passive radar measures any disturbance, and nearly everything disturbs radio waves in some frequency range.

Why is the Astera residency the right fit for this work? Why isn’t this problem likely to be adequately addressed (at this stage) by academia, government, or industry?

The Astera residency is especially aligned to the public goods mission of the project compared to other funders, and Offworld Labs is best placed to build the system.

Academia: Research in passive radar started here, but incentives for incremental output discourage big, fast moves. A network's value is the square of the nodes, so non-linear, and the scale needed to be revolutionary is out of scope for academia.

Government: National radar monitoring has existed for decades, but without security clearance, this data is not accessible. 70 years of closed radar data demonstrates this.

Industry: ADS-B flight tracker networks, the closest analog, all started as open-source. There's interest in passive radar commercially, but for defense-tech applications that are necessarily proprietary. Our open-source approach is advantageous because the export control law ITAR, which restricts passive radar, has carve-outs for work in the public domain.

Which parts of your vision are in scope for a 12-18 month residency, and which are intentionally out of scope?

In Scope:

-Deploying ~1000 nodes across the continental US
-Upgrading server infrastructure to handle this level of data
-Smoothen user experience and installation workflow
-Calibrate and validate measurements with known aircraft positions
-Serve a global map and data api of aircraft detections
-Make historical datasets available to download
-Demonstrate proof of concept for at least one additional sensor modality (e.g., meteor, atmospheric, or wildlife sensing)

Out of Scope:

-Companion mobile app for map display and multi-modal validation (i.e., corroborating smartphone sensor data)
-Network-wide sensor modalities beyond aircraft detection
-In-house neural net training

We will start with aircraft detection because that is the most feasible place to start. Commercial aircraft have large radar cross-sections, are ubiquitous in the environment, and broadcast independent telemetry for cross-referencing.

How do you plan to approach producing open, public-domain outputs during the residency, and what considerations or tradeoffs do you anticipate?

All code and documentation are open-source, all hardware is built from commercial-off-the-shelf parts, and all data is made freely available. All code is available on GitHub, data is streamable through our API, and automated bulk data requests are hosted on Backblaze and served through a Cloudflare frontend.

We are actively soliciting collaborators to use and validate our datasets and to demonstrate their usefulness. A collaboration with the University of British Columbia to train models with the data is now in the early stages.

No commercial investment has been taken in order to maintain this approach, which has been a tradeoff. However, we have been able to raise philanthropic funds, allowing partnerships and goodwill that would not be possible if we were a proprietary commercial project. As mentioned, there are regulatory advantages to being open-source, and to comply with these requirements, we must always make code and data public.