Smart mobility becomes paramount for meeting net-zero targets. However, autonomous, self-driving and electric vehicles require more than ever before an efficient, resilient and trustworthy computational offloading backbone that expands throughout the edge-to-cloud continuum. Utilizing on-demand heterogeneous computational resources for smart mobility is challenging and often cost-ineffective. This paper introduces SMOTEC, a novel open-source testbed for adaptive smart mobility experimentation with edge computing. SMOTEC provides for the first time a modular end-to-end instrumentation for prototyping and optimizing placement of intelligence services on edge devices such as augmented reality and real-time traffic monitoring. SMOTEC supports a plug-and-play Docker container integration of the SUMO simulator for urban mobility, Raspberry Pi edge devices communicating via ZeroMQ and EPOS for an AI-based decentralized load balancing across edge-to-cloud. All components are orchestrated by the K3s lightweight Kubernetes. A proof-of-concept of self-optimized service placements for traffic monitoring from Munich demonstrates in practice the applicability and cost-effectiveness of SMOTEC.

翻译：智能出行对于实现净零排放目标至关重要。然而，自动驾驶、无人驾驶和电动汽车比以往任何时候都更需要一个高效、弹性且可信的计算卸载骨干网络，该网络需贯穿从边缘到云端的连续体。利用按需异构计算资源来实现智能出行既具有挑战性，往往也成本低效。本文介绍了SMOTEC，一个面向自适应智能出行实验的新型开源边缘计算测试床。SMOTEC首次提供了一套模块化的端到端工具，用于原型设计和优化增强现实、实时交通监控等智能服务在边缘设备上的部署位置。SMOTEC支持通过即插即用的Docker容器集成SUMO城市交通模拟器、通过ZeroMQ和EPOS进行通信的树莓派边缘设备，以及基于人工智能的边缘到云端去中心化负载均衡。所有组件均由轻量级K3s Kubernetes编排。来自慕尼黑的交通监控自优化服务部署概念验证，在实践中展示了SMOTEC的适用性和成本效益。