Node
Sensor
An STM32-class microcontroller reads the sensors fitted for that site and stays asleep between readings, which is what makes the power budget work.
Technology
Built for places with no power and no signal.
Remote monitoring usually fails on one of two things: getting data out of deep cover, or keeping hardware alive without a service visit. Nature Guards is designed around both constraints.
The signal path
Radio
LoRa
Long range at very low power, reaching kilometres through terrain that defeats Wi-Fi and where cellular coverage does not exist.
Bridge
Gateway
LoRa traffic is converted to Ethernet, putting the whole network's readings onto standard IP infrastructure.
Decision
AI + monitors
AI analyses readings for trends and anomalies across the network; human monitors judge what deserves a response on the ground.
The node, and the node in a tree
Not a render of a plan. The unit on the right ran for six months strapped to a trunk in the Stepanavan pilot, through an Armenian winter, powered only by the panel on its face.
Why LoRa
The alternatives do not survive a forest
Cellular needs coverage that rarely reaches into managed forest, and the radios draw far too much power for a node expected to last years. Wi-Fi and mesh protocols lack the range once trees and terrain are in the way. Satellite works but costs per message and per node in a way that does not scale to hundreds of sensors.
What LoRa gives up, and why it does not matter
LoRa's bandwidth is tiny. That is irrelevant here: an alert is a few bytes — which sensor, what reading, when. Trading throughput for range and power draw is exactly the right trade for this problem.
Sensors available
Nodes are configured per deployment. A dry-season fire-risk site and an anti-logging boundary draw on different subsets of the same family.
Acoustic
Saw noise, engine noise, human activity
Infrared
Flame and hotspot signatures
Heat
Abnormal heat rise against baseline
Temperature
Ambient trend, frost risk, fire conditions
Humidity
Soil and air dryness — irrigation need, and fire load
Motion
Movement of people, vehicles and machinery
Deployment characteristics
- 9nodes per km²
- Nine nodes per square kilometre — 23 per square mile — on 333 m spacing, each detecting within a 150 m radius. Fewer nodes keeps hardware cost and installation effort predictable as a deployment scales.
- 150 m – 1 kmrange per node
- Each node covers a minimum radius of 150 metres in difficult terrain, extending to a full kilometre across flat ground — which is what sets how many nodes a given site needs.
- Solarno battery swaps
- Nodes power themselves from sunlight and recharge indefinitely, so there is no replacement cycle to schedule — removing the service visits that dominate the lifetime cost of any remote sensing deployment.
- Real timenot retrospective
- Detection to alert is a matter of seconds over LoRa, so a response can be mounted while intervention still changes the outcome.
Stack
Hardware
- STM32 microcontrollers
- LoRa long-range radio
- Solar power — no battery replacement cycle
Software
- C++ (firmware)
- Java (platform services)
- AI-assisted trend analysis
Industries
- Agriculture and Forestry
- Environmental Conservation
Planning a deployment, or want the detail behind these figures? Get in touch.