Serverless edge computing provides a lightweight and easily scalable new paradigm for edge computing, which is widespread in many fields. However, its characteristics of fine-grained tasks, short startup times, and fast execution speed bring new challenges in task offloading and latency reduction. In this paper, we consider the task offloading problem of serverless functions in a multi-edge-to-cloud environment. A new hybrid offloading algorithm, Average latency constrained Independent task Hybrid Offloading (AIHO), is proposed aimed at reducing latency so as to enhance serverless computing in edge environment. AIHO integrates a serverless-based three-layer system framework, enabling strategic deployment of serverless functions closer to end devices, and includes four critical components: offloading decision, task sorting, path selection, and function replacement. The proposed algorithm is evaluated by comparing it to other three baselines for similar problems on the same datasets. By focusing more on horizontal collaboration among edge nodes based on the multi-hop communication mechanism, AIHO exhibits higher performance than other baselines. Experimental results demonstrate that it can significantly reduce average latency, optimize resource usage, and enhance overall resilience and efficiency of edge computing systems, marking a substantial advancement in serverless and edge computing integration.

中文翻译：

---

AIHO：增强无服务器多边缘到云系统中的任务卸载并减少延迟


无服务器边缘计算为边缘计算提供了一种轻量级且易于扩展的新范式，该范式在许多领域得到广泛应用。但其任务细粒度、启动时间短、执行速度快等特点给任务卸载和降低延迟带来了新的挑战。在本文中，我们考虑了多边缘到云环境中无服务器函数的任务卸载问题。提出了一种新的混合卸载算法，平均延迟约束独立任务混合卸载 （AIHO），旨在减少延迟，从而增强边缘环境中的无服务器计算。AIHO 集成了基于 Serverless 的三层系统框架，使 Serverless 功能战略性地部署在更靠近终端设备的位置，包括卸载决策、任务排序、路径选择和函数替换四个关键组件。通过将所提出的算法与相同数据集上类似问题的其他三个基线进行比较来评估该算法。AIHO 基于多跳通信机制，更注重边缘节点之间的横向协作，表现出比其他基线更高的性能。实验结果表明，它可以显著降低平均延迟，优化资源使用，并提高边缘计算系统的整体弹性和效率，标志着无服务器和边缘计算集成取得了重大进步。