Extreme Edge Sensing-as-a-Service: Bridging Containerization for IoT End Devices

On the extreme edge of the IoT deployment, micro-controllers are often managed as semi-static software environments, with restricted multitenancy capabilities. This limitation reduces potential monetization opportunities for the IoT infras-tructure owners and restricts the flexibility for external clients to define custom sensing or edge processing strategies. In this paper, we introduce a radical shift in vision through the new paradigm of Extreme Edge-Sensing-as-a-Service, which allows clients to securely and efficiently rent sensing, computation, and communication resources from third-party IoT end devices. We support this new concept with the design and implementation of the FleXE framework that leverages research contributions in IoT security, edge computing and resource allocation algorithms. The core component of FleXE is a novel, lightweight containerization mechanism for embedded systems with secure access to shared sensing peripherals based on Berkeley Packet Filter (BPF). Furthermore, we incorporate this contribution with QoS-based orchestration algorithms designed to allocate containers to IoT end devices seamlessly for clients, while maximizing revenue for owners. Experimental results demonstrate that our containerization solution introduces negligible overhead compared to bare-metal or pure-Kernel (e.g., Zephyr) solutions, while also effectively handling secure sensing operations. Finally, we demonstrated the efficiency of the proposed container allocation heuristic through extensive simulations.