Building a Sustainable Edge Monitoring System for Coral Reefs Using k0s, NATS, and Raspberry Pi Clusters
)
Challenge: Monitor and protect coral reefs in remote coastal waters
In tropical marine environments, coral reefs are among the most diverse and fragile ecosystems. The health of coral and associated algae are directly influenced by various environmental conditions, such as:
Temperature
pH and Carbonate Chemistry
Salinity
Light Intensity
Nutrient Levels
Dissolved oxygen
Turbidity / Water Clarity
Water Flow / Currents
Arkatech NGO is a Vietnam-based nonprofit that leverages technology to support ocean conservation. The organization needed to monitor these critical environmental variables around the clock in order to ensure optimal growing conditions for corals and associated algae in Vietnam and Mauritius. Adding to the complexity, measurements needed to be taken approximately 150 meters offshore. Deploying robust monitoring infrastructure in open water posed challenges such as unstable connectivity, lack of power, and difficulty in maintaining physical access. Their initial edge computing setup proved inefficient due to high licensing costs, vendor lock-in, and limited customization.
Solution: The Submersible Hardware Stack
To address these unique constraints, the team developed a custom edge computing solution designed for sustainability, autonomy, and resilience in harsh conditions. The hardware stack consisted of a submersible buoy and enclosed in an IP68 waterproof enclosure, housing a Raspberry Pi 4 as the compute node. Power was supplied through an 18W curved solar panel, charging Lithium Polycarbonate batteries to keep the unit operational day and night.
The team integrated a suite of oceanographic sensors to collect scientific data, and a 3G module with an antenna provided connectivity. In a second-generation setup, the team further enhanced the system with cameras and pumps to enable advanced observation and potential interventions, such as localized nutrient dispersal.
Solution: The Edge Software Stack (k0s + NATS + Golang App)
The core of the system’s orchestration layer was k0s, an open source, CNCF sandbox Kubernetes distribution originated by Mirantis. Designed as a single binary with no external dependencies, k0s is compact and self-contained, making it ideal for the project's resource-constrained edge environments. k0s runs reliably on Raspberry Pi and offers compatibility with upstream Kubernetes and full support for high availability. Mirantis Senior DevOps Engineer Prashant Ramhit played a key role in deploying the k0s clusters and application testing. The k0s streamlined architecture made provisioning and deployment simple, and allowed applications to be self healing through features like:
Auto-restart
Health checks
Pod rescheduling
Resource Limits
Installation takes just three commands, making it easy for even beginners to set up a production-ready Kubernetes environment with minimal effort.
To process and transmit data efficiently, the Arkatech team implemented a Golang-based monitoring application that interfaces with the sensors and connects through Neural Autonomic Transport System (NATS) — an open source, lightweight, high-performance messaging system for distributed systems, IoT devices, and cloud native applications. NATS enables fast, reliable communication between devices and the back-end infrastructure, even in conditions of intermittent connectivity.
The solution collects, processes, and transmits data in real time, allowing scientists to monitor changes in the reef ecosystem and respond quickly to adverse conditions.
Below is an illustration of the solution architecture:
This innovative setup embodies the concept of “Kubernetes at the edge.” It runs full Kubernetes clusters in self-contained, solar-powered units, floating on or below the surface of the ocean. It turns what was essentially a plastic enclosure with a tiny motherboard into an autonomous, intelligent node capable of orchestrating workloads, maintaining uptime, and collecting mission-critical data.
The solution not only reduces hardware and software costs, but also eliminates vendor lock-in, giving the team full control over customization and updates. Thanks to the open nature of k0s, updates and patches can be deployed quickly, with minimal downtime.
Expanding the vision: AI/ML workloads and multi-cloud control with k0rdent
As Arkatech deploys more submersible buoys across more locations, they may need a multi-cluster management system to streamline operations, especially if they decide to integrate AI/ML technologies. With k0s, this can be implemented readily with k0rdent, an open source Distributed Container Management Environment (DCME) designed to rapidly provision workloads across multi-cluster environments. k0s is a foundational component of k0rdent, which supports high-performance computing scenarios, including machine learning and AI workloads, and is engineered to deploy applications quickly with production-grade resilience.
Broader Impact
Arkatech NGO’s real-world implementation of edge computing — leveraging open source technologies like k0s and NATS running on readily-accessible Raspberry Pi devices — offers a compelling model for future environmental and edge applications. It shows how Kubernetes can run far away from the data center and thrive even when power, space, and connectivity are limited.
The project successfully delivered:
A fully autonomous marine monitoring platform
Real-time insights into reef health
Ultra-low-cost, solar-powered edge deployments
An open, customizable framework adaptable to other fields
Explore More
Learn more about k0s and how it enables edge-native Kubernetes deployments:
)
)
){kind=logo}
)
)
