The Internet of Things (IoT) has led to a surge in smart devices, generating vast volumes of data. Cloud computing offers scalability but does not suffice for many real-time and privacy-sensitive IoT applications. This limitation has prompted a blend of both edge and cloud resources, creating the need for seamless integration, known as the “compute continuum“. Testing applications and resource management techniques within this continuum is vital but can be very complex. Simulation and emulation are preferred methods, with emulation providing more accurate representations of real-world environments. In this paper, we introduce iContinuum, a novel emulation toolkit facilitating an intent-based platform for edge-to-cloud testing and experimentation. Leveraging Software-Defined Networking (SDN) and containerization, iContinuum enables experimentation and performance evaluation while aligning application requirements with actual performance. We present our detailed architecture, implementation, and evaluation of iContinuum, showcasing how our proposed toolkit bridges the gap between simulation and real-world deployment within compute continuum environments, and further demonstrate the effectiveness of Intent-Based Scheduling through a specific use case.

Serverless edge systems simplify the deployment of real-time AI-based Internet of Things (IoT) applications at the edge. However, the heterogeneity of edge computing nodes – in terms of both hardware and software – makes load balancing challenging in these systems. In this paper, we propose a performance-driven, empirical weight-tuning approach to achieve effective load balancing based on the characteristics and capabilities of the nodes. By extensively profiling the nodes, we gather knowledge on performance metrics such as throughput, energy efficiency, response time, AI accuracy, and cost. Using this acquired knowledge, we introduce a weighted round-robin strategy to optimize the performance metrics according to their observed significance. To address multiple objectives, we introduce a multi-objective method that aims to strike a balance between any arbitrary set of performance objectives simultaneously. Additionally, we explore a coordinated distributed approach to overcome the limitations of centralized load balancing. Next, we introduce Hedgi, a heterogeneous serverless edge architecture designed to efficiently configure and utilize the derived load balancing policies, validated empirically. To demonstrate the practicality of Hedgi, we containerize and serverlessize a real-time object detection application. Extensive empirical studies are conducted using Hedgi to evaluate the performance of the proposed load balancing approach. The results provide valuable insights into the design trade-offs of various load balancing policies and system designs in the heterogeneous serverless edge.

Serverless edge computing is a specialized system design tailored for Internet of Things (IoT) applications. It leverages serverless computing to minimize operational management and enhance resource efficiency, and utilizes the concept of edge computing to allow code execution near the data sources. However, edge devices powered by renewable energy face challenges due to energy input variability, resulting in imbalances in their operational availability. As a result, high-powered nodes may waste excess energy, while lowpowered nodes may frequently experience unavailability, impacting system sustainability. Addressing this issue requires energy-aware resource schedulers, but existing cloud-native serverless frameworks are energy-agnostic. To overcome this, we propose an energyaware scheduler for sustainable serverless edge systems. We introduce a reference architecture for such systems and formally model energy-aware resource scheduling, treating the function-to-node assignment as an imbalanced energy-minimizing assignment problem. We then design an optimal offline algorithm and propose faasHouse, an online energy-aware scheduling algorithm that utilizes resource sharing through computation offloading. Lastly, we evaluate faasHouse against benchmark algorithms using real-world renewable energy traces and a practical cluster of single-board computers managed by Kubernetes. Our experimental results demonstrate significant improvements in balanced operational availability (by 46%) and throughput (by 44%) compared to the Kubernetes scheduler.

In this paper, we present energy-aware scheduling for Serverless edge computing. Energy awareness is critical since edge nodes, in many Internet of Things (IoT) domains, are meant to be powered by renewable energy sources that are variable, making low-powered and/or overloaded (bottleneck) nodes unavailable and not operating their services. This awareness is also required since energy challenges have not been previously addressed by Serverless, largely due to its origin in cloud computing. To achieve this, we formally model an energy-aware resource scheduling problem in Serverless edge computing, given a cluster of battery-operated and renewable-energy powered nodes. Then, we devise zone-oriented and priority-based algorithms to improve the operational availability of bottleneck nodes. As assets, our algorithm coins terms “sticky offloading” and “warm scheduling” in the interest of the Quality of Service (QoS). We evaluate our proposal against well-known benchmarks using real-world implementations on a cluster of Raspberry Pis enabled with container orchestration, Kubernetes, and Serverless computing, OpenFaaS, where edge nodes are powered by real-world solar irradiation. Experimental results achieve significant improvements, up to 33%, in helping bottleneck node's operational availability while preserving the QoS. With energy awareness, now Serverless can unconditionally offer its resource efficiency and portability at the edge.