EdgeKeeper: Resilient and Lightweight Coordination for Mobile Edge Computing Systems
Institutional Repository Document
Mobile Edge Computing (MEC) has been gaining significant interest from first responders and tactical teams, primarily because they can employ handheld mobile devices to form a computing cluster (for computing tasks like face/scene recognition, virtual assistance) when connectivity to the cloud is not present or it is limited. High user mobility in first responder or tactical environments makes MEC challenging, as wireless links observe substantial fluctuations. Typical cloud-based coordination (e.g., ZooKeeper-based service discovery and coordination, device naming, security) needed by edge computing tasks cannot work in these environments. Driven by the need for a resilient and lightweight coordination service, in this paper, we design and implement ek to provide cloud-like coordination for MEC systems. It provides naming, network management, application coordination, and security to distributed edge computing applications. It maintains an edge cluster among devices and intelligently stores its data on a group of replicas to guard against node failure and disconnections. We provide a full-system implementation of EdgeKeeper for Android and Linux platforms. We have integrated EdgeKeeper with existing MEC applications and performed real-world performance evaluations in a wide-area search and rescue operation conducted by first responders, which proves it to be lightweight and suitable for mobile devices.