Latest / The Edge Computing Podcast with Fexingo: Local Compute, CDNs, and Distributed Infrastructure / How Edge Computing Is Making Real-Time Disaster Response Possible
Transcript
- Lucas: When Hurricane Maria hit Puerto Rico in September 2017, it knocked out 95 percent of the island's cell towers. For most people, that meant no way to call for help, no way to coordinate relief, no way to know if their family was safe for weeks or even months. Luna: And that's exactly where edge computing steps in — not as a convenience, but as a literal lifeline. Lucas: Exactly. So today I want to talk about how edge computing is making real-time disaster response possible. The case I'm going to focus on is Project Owl — o w l — which stands for 'Organization for World Learning' and their mesh-network system that got deployed on the ground in Puerto Rico. Luna: I remember reading about that. They used these little nodes that basically created a local Wi-Fi mesh, right? Lucas: Yeah. Each node is about the size of a lunchbox — it's got a battery, a small compute module, and a cognitive radio that can dynamically switch frequencies to find a clear channel. The key insight is that these nodes don't need a central server or a working cell tower. They form a peer to peer network that routes data locally, node to node, until it reaches a gateway that can connect to the internet via satellite or a surviving fiber link. Luna: So it's edge computing because the processing happens on the node itself, not in some distant cloud data center that might also be down. Lucas: Right. And that's the fundamental shift. Traditional disaster communications rely on centralized infrastructure — cell towers, fiber backhauls, cloud servers — which are exactly the things that get destroyed in a hurricane or an earthquake. Edge nodes flip that model. They bring compute and connectivity to the very edge of the network, right where people are. Luna: What kind of compute are we talking about? Are these Raspberry Pi-level devices, or something beefier? Lucas: They're more capable than a Pi but not a full server. The Project Owl nodes use an Intel Atom processor, about 4 gigabytes of RAM, and a software-defined radio. So enough to run a local web server, handle mesh routing, and do basic data processing — like compressing images or translating voice messages — but nothing that requires a GPU cluster. The point is to be low-power and rugged. Each node runs for about 8 to 12 hours on a battery pack, and you can recharge it from a car battery or a small solar panel. Luna: So how fast can you actually get one of these networks up and running after a disaster? Lucas: That's the remarkable part. In Puerto Rico, Project Owl's first deployment team got a 10-node mesh network operational within 8 hours of arriving on the ground. Compare that to the traditional cell tower recovery timeline — the FCC reported that even 11 months after Maria, about 17 percent of cell sites were still not back online. So we're talking about a difference between hours and months. Luna: That's an insane gap. And I imagine the edge nodes also help with coordinating the relief itself — like, not just texting family but actually helping FEMA or local emergency services prioritize where to send supplies. Lucas: Exactly. One of the applications that emerged from Puerto Rico was a simple triage form that ran on the mesh. First responders could enter patient status, location, and supply needs on a tablet, and that data would propagate across the mesh to a central coordination node. Because the processing was local, the system stayed responsive even when the satellite uplink was slow or intermittent. That's the edge advantage — you don't need a constant connection to the backbone to keep working. Luna: It reminds me of how some of these same principles are now being built into public safety networks. I know the FirstNet initiative in the US uses some edge-based prioritization, but it's still largely dependent on cell towers. Lucas: Yeah, FirstNet is a big step forward — it gives first responders dedicated bandwidth on the existing cellular network. But as Maria showed, if the towers themselves are gone, dedicated bandwidth doesn't help you. That's where mesh networks with local compute are fundamentally different. They're designed to work when everything else is offline. Luna: You know, this conversation is a great reminder of why we keep this podcast ad-free. It lets us dive into stories like this one without worrying about pleasing a sponsor. If today's episode gave you a new perspective on how technology can really help people, and you want to support that choice, the link is buy me a coffee dot com slash fexingo. No pressure, just a way for the people who find value in this to keep it going. Lucas: Yeah, I really appreciate that. And it's true — being ad-free means we can follow the thread wherever it goes, even if it's a small nonprofit project instead of a billion-dollar company. So, back to the mesh: one of the other things Project Owl did was build a local text-messaging app that worked entirely on the edge nodes. Luna: Right, because SMS requires a working cellular network, and WhatsApp or iMessage need internet. But a local web app on the mesh can let people send messages without either. Lucas: Exactly. And they made it dead simple — anyone with a phone that has Wi-Fi could connect to the node's SSID, open a browser, and start messaging. No app install, no account creation. That's the edge computing design philosophy: meet users where they are, with the devices they already have, and do the processing as close to them as possible. Luna: What about data security? In a disaster scenario, you're transmitting medical info, location data, maybe even financial aid details. Is edge computing inherently more secure, or does it introduce new risks? Lucas: It's a mixed bag. On one hand, because data doesn't travel all the way to a central cloud, there's less exposure to interception over long-haul links. And the mesh is decentralized, so there's no single point of failure or single target for a cyberattack. On the other hand, each node is physically accessible — someone could steal a node and try to extract data from it. Project Owl encrypted all data at rest on the nodes, and they designed the system so that if a node is compromised, the rest of the mesh can isolate it. Luna: So the edge architecture actually strengthens resilience against both natural and human threats. That's a powerful combination. Lucas: It is. And we're starting to see this approach formalized. The FCC's Communications Security, Reliability, and Interoperability Council has actually recommended that local governments and utilities deploy edge-based mesh networks as a supplement to primary cellular infrastructure. Some counties in California, after the 2018 wildfires, have started pilot programs with nodes placed on fire stations and schools. Luna: Are there any other disaster scenarios where edge computing is making a real difference right now? Lucas: Yes — earthquake early warning systems. In Japan, the system uses a network of seismometers that send data to a central processing center. But that introduces latency. Researchers at UC Berkeley have been testing an edge-based approach where each seismometer has a small compute module running a machine learning model that can detect P-waves — the primary, less destructive waves — and issue a local alert within milliseconds, before the S-waves arrive. Luna: So instead of waiting for the cloud to confirm an earthquake, the sensor itself decides and broadcasts the warning locally. That could buy seconds — which in a quake is huge. Lucas: Exactly. The model is trained to distinguish between actual seismic activity and noise from trucks or construction. It runs on a low-power edge device that costs around 200 dollars. In a test deployment along the Hayward Fault in California, the system achieved a median alert time of 2.1 seconds after the P-wave detection — that's about 1.5 seconds faster than the centralized network. Luna: Two seconds might not sound like much, but for someone dropping to the ground or for an automated gas valve shutoff, it's everything. Lucas: Right. And the edge approach also scales better. You don't need to build a massive centralized data center to handle thousands of sensors. Each node is self-contained. You can add more sensors without bottlenecking the network. It's the same kind of distributed resilience we saw with Project Owl. Luna: So what's the biggest barrier to wider adoption? Is it cost, or awareness, or maybe the fact that emergency response systems are slow to change? Lucas: I think it's a combination of all three. The hardware cost has come down dramatically — a Project Owl node is about 1,200 dollars, and an earthquake sensor around 200. But you still need training, maintenance, and integration with existing 911 systems. Plus, there's the inertia of traditional telecom: carriers are built around centralized infrastructure, and mesh networks are a fundamentally different business model. They don't generate monthly subscription revenue. Luna: That's a really important point. If we're relying on the market to solve this, the incentive structure is misaligned. The people who benefit most — the victims of a disaster — aren't the ones paying for the infrastructure. Lucas: Exactly. That's why a lot of this is being driven by nonprofits, academic labs, and forward-thinking local governments. But there are signs of commercial interest too. Companies like goTenna and Sonim are making consumer mesh devices for outdoor recreation, and the tech is filtering up. If the outdoor market grows, it could bring down component costs and make disaster-grade nodes even cheaper. Luna: It's a classic edge computing story: take compute and intelligence out of the centralized cloud, push it to where the data is generated, and suddenly you unlock capabilities that were impossible before. Lucas: Yeah. And when the data is a cry for help, or a seismic warning, or a request for medical supplies, the latency savings aren't just a performance metric. They're the difference between life and death. That's the real promise of edge computing in disaster response — not faster Netflix streams, but faster connection to the people who need it most. Luna: So what's next? Are there specific regions or projects you're watching? Lucas: I'm keeping an eye on the Pacific Northwest. The Cascadia Subduction Zone is overdue for a major earthquake, and there's a coalition of universities and emergency management agencies working on a multi-year pilot to deploy edge-based early warning and mesh communications along the coast. If that proves out, it could become a model for other regions. Also, Project Owl is expanding their work to wildfire-prone areas in Australia. So I think we'll see a lot more of this in the next 3 to 5 years. Luna: That's good to hear. Because right now, most of us take connectivity for granted — until it's gone. And then we realize how fragile the centralized model really is. Lucas: Exactly. The edge isn't just about speed or efficiency. It's about building systems that don't have a single point of failure. And in a world where disasters are becoming more frequent and more intense, that kind of resilience is exactly what we need.