At SAC LABs we developed and deployed a 5 year edge computing technology demonstrator in harsh northern environments using off the shelf ad custom designed components. Combining onsite edge computing for data analytics, 4G/LTE, remote sensing hardware and material sciences to protect the electronics from 80 degree temperature swings.

We combined various IoT environmental modules for remote sensing and data collection applications.  We used RaspberryPi 4 and various hardware systems for onsite compute, while relying on Ubuntu/NodeRed stack for data collection.  The edge compute collector then sent the data back to the cloud datacenter, where a MongoDB cluster stored the data for analyst to process.

These solutions form a designing complex that can be employed to remote locations that cannot be accessed onsite frequently. They can be deployed broadly over a large geographic area and interconnected through WiFi, LTE of Satellite uplinks to provide broad geographic data collection.

The underlying software stack can be managed using DevOps CI/CD process workflows to deploy new collection capabilities from a central location to the edge without onsite intervention.

DevOps workflow for SAC modules:

• Automation: Automate workflows, code testing, and infrastructure provisioning to cut down on overwork.
• Iteration: Using small chunks of proven code during sprints to support the speed and frequency of deployments.
• Continuous improvement: Continuously test and learn from previous failures in order to act on feedback to optimize performance, cost, and time to deployment.
• Collaboration: Unite teams and foster communication to maintain high quality.

Redundancy and Resiliency Model

Using separate dedicated modules (ex: compute, sensor, battery and communications) increases reliability, remote diagnostic and self healing capabilities of the SAC module. Building multiple resources that serve the same function allows for the ability to fail-over modules to other modules in the event of the loss of a single primary resource.

Our Resiliency model increases both remote diagnostic intelligence and the longevity of individual SAC modules. By being able to test and diagnose primary systems and read out fault codes from the redundant system, we are able to avoid unnecessary onsite maintenance for low battery or false failed component assumptions. With in-cloud data, we can see when the last data was read out, and the last status of each module. In the event of low solar re-charge in conjunction with low temperatures, we can determine that when weather conditions improve the primary systems will self heal with no immediate need for onsite response.

• Autonomy: Multi-module redundancy, diagnostic and fail-over
• Stateless: Distributed services that can be independently managed
• Observability: Intelligent alerting and monitoring using dedicated tracing and continuous reporting.
• Reliability: High service availability and redundancy, with remote, onsite or automated backup and recovery.