Community-Oriented Edge Computing Platform
Edge Computing , Extreme Edge , EEDs , Resource Allocation , Container Placement
The surging demand for Edge Computing (EC) to cope with the proliferation of latency-critical and data-intensive applications has inspired the concept of recycling ample yet underutilized computational resources of end devices, also referred to as Extreme Edge Devices (EEDs). However, maintaining data privacy and cost efficiency remain core challenges for the viability of EED-enabled computing paradigms. In this thesis, we propose the Community Edge Platform (CEP). CEP exploits business, institutional, and social relationships to build clusters and communities of requesters and EEDs to eliminate recruitment barriers and preserve privacy. Furthermore, CEP utilizes a hierarchical control paradigm to prioritize the enrollment of nearby devices as workers. CEP first seeks workers within the requester's cluster, which is composed of devices owned by the same user. In case of a lack of a suitable device, CEP resorts to devices in other clusters that belong to the requester's community. In addition to the underlying architecture and structure of CEP, we also consider the fact that community-imposed constraints on resource allocation can lead to unbalanced work distribution. To address this issue, we introduce the Community-Oriented Resource Allocation (CORA) scheme. CORA strives to resolve community restrictions, minimize the flowtime and makespan for the allocated services, while retaining a reasonable scheduler runtime for real-time resource allocation. Towards that end, we formulate the resource allocation problem as a Bipartite Graph Matching problem. Furthermore, we expose tuneable parameters that allow prioritizing flowtime or makespan, which makes CORA suitable for a wide variety of scenarios. A detailed comparative study demonstrates the leverage of CEP compared to 12 prominent edge computing platforms. Moreover, extensive simulations show that CORA outperforms six prominent heuristic-based resource allocation schemes by up to 28% and 6% in terms of average makespan and flowtime, respectively, while sustaining an adequate level of runtime.