Latest / The Edge Computing Podcast with Fexingo: Local Compute, CDNs, and Distributed Infrastructure / How Edge Computing Powers Real-Time Energy Trading on Local Microgrids
Transcript
- Lucas: So earlier this week I was reading about a pilot project in Brooklyn — the Brooklyn Microgrid — where homes with rooftop solar panels are selling excess energy to their neighbors in real time, all coordinated through edge computing nodes. Luna: Wait — like a local energy stock exchange? You're buying and selling kilowatt-hours the way you'd trade shares? Lucas: Exactly. And it's not a future concept — it's been running since 2016 in the Gowanus and Park Slope neighborhoods. About sixty homes and small businesses participate. They each have a smart meter and an edge computing device about the size of a cable modem. Luna: And that edge device is doing what, exactly? Lucas: It's running a local instance of a distributed ledger — essentially a blockchain — but specifically tuned for energy transactions. Every few seconds, it sends a bid or ask based on current solar generation and household consumption. The edge nodes in the neighborhood collectively validate and settle trades — peer to peer — without ever touching a centralized utility server. Luna: So the edge is acting as both the trading platform and the clearinghouse. That's a pretty big departure from how utilities normally work. Lucas: It's a fundamental shift. In a traditional grid, power flows one way — from a central plant to your home. Here, the grid is bidirectional, and the transactions need to happen in near-real time because solar output is volatile. A cloud passes overhead and generation drops by forty percent in seconds. The local balancing requires sub-second decision making. Luna: And that's where the edge latency advantage really kicks in. Sending a trade request to a cloud server fifty miles away and back could take a hundred milliseconds or more. By that point, the cloud has already passed and the grid condition has changed. Lucas: Right. The Brooklyn Microgrid pilot uses what they call 'transactive energy' — each edge node runs an algorithm that matches supply and demand every five seconds. That's orders of magnitude faster than what a central utility SCADA system can do, which typically polls meters every fifteen minutes. Luna: So the edge isn't just faster — it's enabling a completely different market structure. A fifteen-minute settlement window wouldn't work for peer to peer trading because the price of solar power changes with the weather. Lucas: Exactly. And that's the key insight. The edge makes it possible to have granular, real-time pricing that reflects actual grid conditions. In the pilot, prices can range from about two cents per kilowatt-hour on a sunny afternoon to twelve cents on a cloudy evening. That dynamic pricing incentivizes homeowners to shift their usage — run the dishwasher when the sun is shining and power is cheap. Luna: I wonder about the hardware side. What kind of edge devices are we talking about? Are they purpose-built for energy trading, or is it more like a Raspberry Pi running custom software? Lucas: In the Brooklyn pilot, they used a device from a company called LO3 Energy — a small Linux box with a cellular modem and a local radio for communicating with the smart meter. It's basically a ruggedized single-board computer. But the bigger challenge is standardization. If you want to scale this to millions of homes, you need a common protocol that works across different manufacturers' hardware. Luna: And that's where the industry is still early. There's no universal standard for edge-based energy trading yet. But I know the IEEE is working on a standard for transactive energy — IEEE 2030.5 — which touches on some of this. Lucas: Good point. And there are other pilots too — in Australia, in the Netherlands, in parts of California. But the Brooklyn Microgrid is probably the most documented. They published a lot of data on transaction latency and reliability. One stat I saw: average settlement time is under two seconds, including the blockchain consensus step. Luna: Two seconds is still a long time in grid terms. If you have a sudden imbalance — say a large load switches on — the frequency can dip dangerously in milliseconds. How do they handle that? Lucas: That's the big question, and it's why edge-based trading isn't replacing primary grid regulation — it's sitting on top of it. The local utility still has fast frequency response reserves, like batteries and spinning generators. The edge market handles the slower, economic dispatch layer — the trades between prosumers. Think of it as a marketplace that optimizes within the safety envelope that the utility maintains. Luna: So the edge is doing the 'smart' coordination, but the physical safety net remains centralized. That makes sense. But it also means the edge nodes have to be incredibly reliable — if they miss a heartbeat or a trade confirmation, the grid could destabilize. Lucas: Absolutely. That's why every node in the Brooklyn pilot has battery backup and a cellular failover. They also replicate the ledger across multiple nodes, so if one goes down, the others can continue settling trades. The system is designed to degrade gracefully — if too many nodes are offline, it falls back to a simpler 'fixed price' mode. Luna: I want to talk about the economics for a moment. How much money are these homeowners actually making? Is this a meaningful revenue stream or more of a hobby? Lucas: In the pilot, the average household earned about three hundred to five hundred dollars per year from selling excess solar power. That's not going to pay off a solar installation on its own, but it does improve the payback period by maybe fifteen to twenty percent. The bigger value might be in reduced grid infrastructure costs — if you can balance locally, you don't need to upgrade transmission lines as often. Luna: And that's where the utility actually benefits. They can defer capital expenditure on substations and transformers. So there's a business case for them to support this kind of edge market, even if it means losing some control. Lucas: Right. And some utilities are starting to experiment. In California, a program called 'Energy Imbalance Market' uses edge devices to allow distributed resources to participate in wholesale markets. But that's still centralized compared to the Brooklyn model. Luna: Let's zoom out. We've talked about one specific pilot in one neighborhood. What would it take to scale this to a city-wide or regional level? Lucas: The biggest hurdle isn't technology — it's regulation. Most states still have laws that prevent retail customers from selling electricity to each other. The Brooklyn Microgrid operates under a special exemption from the New York Public Service Commission. So the regulatory framework has to catch up. Luna: And on the tech side, do we need better edge computing hardware, or is the current generation sufficient? Lucas: I think the edge hardware is good enough today. The real bottleneck is interoperability — getting different brands of solar inverters, smart meters, and edge devices to speak the same language. That's a software and standards problem, not a hardware one. Groups like the OpenFog Consortium and the Industrial Internet Consortium have been working on reference architectures. Luna: Speaking of standards, there's also the cybersecurity angle. If every edge node is a potential attack surface for manipulating energy prices or even destabilizing the grid, that's a serious concern. Lucas: Huge concern. In the Brooklyn pilot, they used a permissioned blockchain — only approved nodes can participate. And each node has hardware security modules for cryptographic key storage. But as you scale to thousands of nodes, you need automated certificate management and intrusion detection at the edge. Some researchers are exploring machine learning models that run on the edge devices themselves to detect anomalous trading patterns. Luna: That's a really interesting use case — using edge AI to protect the edge market. It's almost like an immune system for the grid. Lucas: Exactly. And we're already seeing startups like GridBeyond and AutoGrid offering edge-based optimization and security services for distributed energy resources. The ecosystem is growing fast. Luna: I'm curious about the user experience. Does the homeowner need to be actively trading, or is it automated? Lucas: It's fully automated. You set your preferences — minimum price you're willing to sell at, maximum you're willing to buy at — and the edge node executes trades on your behalf. You get a weekly report showing your earnings and carbon savings. Most participants treat it as a passive income stream. Luna: So it's like a robo-advisor for your solar panels. That actually makes it much more accessible. Lucas: And that's the kind of thing we like to dig into on this show — real-world applications where edge computing isn't just a theoretical improvement but a fundamental enabler. And you know, conversations like this take time to research and produce. Luna: They do. And we keep the episodes ad-free because we want you to hear the full story without interruption. If you find value in that — if today's tech conversation gave you something usable — you can support the show. Lucas: The link is buy me a coffee dot com slash fexingo. No pressure, just an option to keep this independent. Now, back to the grid — because there's another layer to this story that I think is even more interesting. Luna: What's that? Lucas: The edge nodes in the Brooklyn pilot are also running a small machine learning model that forecasts local solar generation fifteen minutes ahead. That forecast is used to optimize the trading algorithm — so if the model predicts clouds, the node will try to sell power now rather than wait. Luna: So the edge is doing both the transaction processing and the prediction. That's a pretty good example of why you need local compute — you can't afford to send high-frequency meter data to the cloud for inference. Lucas: Exactly. And that's where I think the real future is. Not just peer to peer trading, but a whole suite of edge-native services — forecasting, fault detection, demand response — all running on the same hardware. The Brooklyn Microgrid is a glimpse of that. Luna: One question I still have: what happens to the edge nodes when the internet goes down? Can they keep trading? Lucas: In the pilot, they can. The nodes form a local mesh network using radio frequencies. They can continue trading within the neighborhood even if the wide-area internet is cut. The trades are stored locally and reconciled with the central ledger once connectivity returns. That's a huge resilience advantage. Luna: That's a great example of edge autonomy. You don't need the cloud for the core function to work. Lucas: And that's the kind of robustness that large-scale microgrids will need. If we're going to depend on distributed energy for a meaningful share of our power, the control system can't have a single point of failure. Edge computing gives us that distributed intelligence. Luna: Well, I'm going to keep an eye on how the regulatory landscape evolves. It feels like the technology is ready, but the rules aren't. Lucas: That's the story across so many edge computing applications. The tech is ahead of the policy. But given the pace of change, I'd expect to see a lot more microgrid markets popping up in the next couple of years. Luna: Alright, that's a good note to end on. Thanks for digging into this with me. Lucas: Always a pleasure. We'll be back next time with another edge case.