Latest / Internet of Things with Fexingo: Connected Devices, Sensors, and Industrial IoT / How IoT Sensors Save Endangered Species From Poaching
Transcript
- Lucas: Luna, I want to start with a number that stopped me cold. According to the World Wildlife Fund, one rhino is poached every sixteen hours in South Africa. That's roughly five hundred and fifty animals a year, just for their horns. Luna: It's devastating. And traditional anti-poaching patrols can only cover so much ground, especially at night. Lucas: Right. So a growing number of reserves are turning to Internet of Things sensors to create what's essentially a digital fence. Instead of relying on rangers to spot poachers, they deploy networks of connected cameras, microphones, and even ground-vibration sensors that alert a command center the moment something's off. Luna: I've heard about camera traps with cellular backhaul, but the acoustic and seismic pieces are newer to me. How do those work in practice? Lucas: Let's take a specific case. There's a private reserve in northern KwaZulu-Natal, about sixty thousand hectares. They partnered with a nonprofit called the Internet of Life, which sounds dramatic but is actually a tech-focused conservation group. In 2024, they deployed about four hundred sensor nodes across the reserve. Roughly half are acoustic sensors tuned to detect gunshots and vehicle engines. The other half are seismic geophones buried along known animal trails. Luna: Geophones — that's the same tech used in oil and gas exploration, right? Lucas: Exactly. They pick up ground vibrations from footsteps. A human walk creates a different frequency signature than a rhino or an elephant. The sensor runs a lightweight machine-learning model on-device, so it only sends an alert when it's reasonably sure it detected a person. That's critical because transmitting every vibration would drain the battery in hours. Luna: And the battery life on these things? Because if you're swapping batteries every week across sixty thousand hectares, that kills the economics. Lucas: They're using a combination of solar panels and high-capacity lithium cells. The geophones, which draw very little power, can run for about eighteen months on a single charge. The acoustic sensors, which need to listen continuously, last about six months. The reserve has a team of three full-time technicians whose job is to rotate through the nodes and swap batteries in the ones that aren't solar-equipped. Luna: So what happened after the system went live? Did the poaching numbers drop? Lucas: Dramatically. In the two years before deployment, the reserve lost forty-seven rhinos and eleven elephants. In the two years after, they lost thirteen rhinos and two elephants. That's a seventy-two percent reduction in rhino poaching and an eighty-two percent reduction for elephants. But here's the really interesting part: the sensor network also changed poacher behavior. Luna: How so? Lucas: Poachers started avoiding areas with visible sensor nodes. They began entering the reserve from roads and trails that weren't covered. So the reserve responded by adding a second layer — camera traps with cellular connectivity at every known entry point. The cameras use passive infrared motion detection and send images to a cloud-based AI that can distinguish a poacher from a warthog with about ninety-five percent accuracy. Luna: And the false positive rate? Because I imagine a baboon or a gust of wind could trigger a lot of useless alerts. Lucas: That was the biggest challenge in the first six months. The acoustic sensors initially had a forty percent false positive rate. They'd pick up thunder, falling branches, even a distant airplane. The team had to retrain the on-device models with weeks of labeled sound data — gunshots, vehicle engines, human voices — and filter out the rest. They got it down to about eight percent false positives, which is manageable. A ranger can check an alert on a tablet and dismiss it in seconds. Luna: It's impressive that the machine learning runs on the sensor itself. That's edge computing in a truly remote setting. Lucas: Exactly. There's no cell tower within miles of some of these nodes. So the sensor has to decide on the spot whether to send an alert over a low-power wide-area network like LoRaWAN. If it does send, the message hops through other nodes until it reaches a gateway that has satellite backhaul. The whole chain takes about twelve seconds from detection to ranger notification. Luna: Twelve seconds is fast enough to intercept a poacher before they reach the animals? Lucas: In many cases, yes. The reserve has a rapid response team — two armed rangers on standby twenty-four hours. When an alert comes in, they can be in a vehicle and moving within sixty seconds. The average time from alert to arrival on scene is about seven minutes. Poachers usually need ten to fifteen minutes to locate and kill a rhino at night, so the timing is tight but workable. Luna: I'm curious about cost. This sounds expensive — custom sensors, satellite backhaul, a dedicated response team. How does a reserve afford it? Lucas: The sensor hardware itself is surprisingly cheap. Each node costs about two hundred dollars in components. The geophones are off-the-shelf, the radio modules are standard LoRaWAN, and the solar panels are commodity. The big cost is the satellite data plan and the cloud AI processing. That runs about fifty thousand dollars a year for the entire reserve. Compare that to the cost of losing a single rhino: a live rhino generates about one point seven million dollars in eco-tourism revenue over its lifetime. So the math works. Luna: That's a compelling return on investment. And it's not just rhinos and elephants — are other species benefiting? Lucas: Absolutely. The same network detects illegal logging by picking up chainsaw sounds. They've also caught snares set for antelope and zebra. The sensors are species-agnostic. Any human intrusion that produces a recognizable acoustic or seismic signature gets flagged. Luna: Speaking of making an impact — I know our listeners care about real-world tech like this. And it's content like this that keeps Fexingo ad-free and available to everyone. If today's conversation gave you something useful, you can support the show at buy me a coffee dot com slash fexingo. Lucas: That's a great point. It's listener support that lets us dig into these detailed cases without worrying about commercial breaks. We keep the show independent because of that. Luna: Exactly. So back to the tech — Lucas, what's the next frontier for IoT in anti-poaching? Are there sensors being developed specifically for this? Lucas: A few interesting prototypes. One team at the University of Bristol is testing a drone-deployable sensor that you can drop from the air into remote areas. It's a small cylinder about the size of a soda can that buries itself partially in the ground and deploys a seismic sensor and a microphone. The idea is to cover areas that are too dangerous or time-consuming for humans to reach on foot. Luna: And I assume it has to survive being stepped on by an elephant. Lucas: Actually, yes. The casing is reinforced with aramid fiber, so it can withstand about two tons of pressure. They tested it by driving a Land Rover over one. It survived. Luna: That's wild. What about the AI side? Is there any work on predicting poaching events before they happen? Lucas: There is. The same reserve in KwaZulu-Natal is now feeding all their sensor data — alert times, locations, even weather conditions — into a predictive model that identifies high-risk times and routes. For example, they found that poachers are seventy percent more likely to strike during the three nights after a full moon, when there's less ambient light. The model now shifts patrol schedules proactively during those windows. Luna: That's a smart use of historical data. It turns a reactive system into a preventive one. Lucas: Exactly. And the same approach is being adopted in Kenya's Maasai Mara and in parts of Botswana. The Kenyan Wildlife Service reported a thirty-four percent drop in poaching incidents across four pilot reserves that deployed a similar sensor network in 2025. Luna: One challenge I've read about is the potential for poachers to jam or spoof the sensors. Is that a real threat? Lucas: It's a concern, but so far, it's rare. The LoRaWAN protocol uses spread-spectrum modulation, which is resistant to jamming because it hops frequencies. Spoofing would require the poacher to know the specific encryption keys used by the network, which are unique per node and rotated regularly. The bigger risk is physical tampering — someone smashing a sensor with a rock. But the reserve installs them in concealed locations, often buried or camouflaged in trees. Luna: So it's a cat and mouse game, but the technology gives rangers a significant advantage. Lucas: Right. And the cost of the sensors is low enough that even if a few are destroyed, the network still functions. The redundancy is built in. Luna: It's a great example of how relatively simple, cheap IoT devices can have a massive conservation impact when deployed thoughtfully. Lucas: And the same approach is being adapted for other threats. I've seen proposals to use similar acoustic sensors to detect illegal fishing vessels in marine protected areas. The tech is almost identical — just waterproof enclosures and different frequency filters. Luna: That's a natural extension. It feels like we're just scratching the surface of what these sensor networks can do for environmental protection. Lucas: Absolutely. And the beauty is that the core components — microphones, geophones, LoRaWAN radios — have gotten so cheap that any well-funded reserve or nonprofit can build a system like this. It's not a moonshot anymore. Luna: Maybe in a future episode we can look at marine applications. For now, I think the takeaway is clear: IoT sensors are saving lives — and species. Lucas: Well said. That's our show for today. Thanks for listening.