Tactical edge – Architecting for Disconnected Edge Computing Scenarios

Tactical edge refers to edge computing environments that support military operations. Such situations are characterized by limited connectivity, high mobility, and the most intense security requirements.

At the tactical edge, the most common needs include deploying, managing, and securing the following technologies:

IoT: Cameras and sensors of every description get cheaper every day. The number of data-gathering devices is growing just as quickly on the battlefield as it is back home. These devices need to either facilitate immediate/local decision-making or move the data they collect to an aggregation point where those things can be done.

Compared to prior decades, when such things would have been done using proprietary protocols and architectures specific to each system, nowadays, the same IoT protocols used in the civilian world are used at the tactical edge.

For example, consider a UAV that drops thousands of LoRaWAN-capable vibration sensors over several square miles of terrain around a Forward-Operating Base (FOB). These could leverage LoRaWAN’s inbuilt geolocation features to triangulate where a threat is coming from without each sensor requiring GPS units of their own. A system like this is so cheap that the unit probably wouldn’t even bother to collect the sensors when they moved on.

ML: One of the most vital roles of a communications officer is to facilitate the rapid movement of information from where the action is to where the intelligence teams can analyze it. Those teams have an ever-increasing amount of data coming at them in the form of audio, video, and IoT sensor data. How can they make sense of it all? How can they know whether a given individual walking by a camera is a known threat or someone to be protected? How do they quickly translate and transcribe thousands of conversations captured by microphones attached to those cameras? Military intelligence teams are relying more and more on ML to perform such tasks, freeing up analysts and enabling them to better advise commanders.

ML models are also being used for the same sorts of things seen in the civilian world. This includes V2X-like monitoring and control of vehicles, condition monitoring of equipment for failure prediction, or optimization of supply chains.

Containers and virtualization: Much like the situation with IoT, the applications needed by military commanders are growing exponentially – and militaries around the world are responding by standardizing on the same tools used in the civilian world to do DevOps, DevSecOps, MLOps, and so on.