A Cloud–Edge Collaborative Enabling Architecture Based on a Heterogeneous Embedded Platform

As the application of artificial intelligence in edge scenarios becomes increasingly widespread, cloud-edge collaborative computing has become a new solution for achieving low latency and reliable intelligent services. However, edge devices are limited by size, energy consumption, and limited computing resources, while tightly coupled embedded platforms lack flexibility and scalability. To address these challenges, this paper proposes a cloud-edge collaborative support architecture based on heterogeneous embedded platforms. At the architectural level, a lightweight cloud-edge framework is designed to support the integration and coordination of heterogeneous edge computing resources. At the hardware level, a modular and multi-domain heterogeneous embedded platform is implemented, which realizes flexible configuration and upgrade of CPUs, NPUs, and FPGAs computing units through standardized interfaces. The proposed architecture can achieve adaptive collaboration among the cloud, edge, and device layers under different network conditions and task requirements. Experimental results show that the proposed solution has improved performance in edge inference, scalability, and operational flexibility, providing a new practical solution for autonomous edge computing clusters and intelligent base station systems.