| With the advancement of smart manufacturing 2025,new requirements have emerged in the field of building control in remote maintenance,device access,flexible resource allocation,low-bandwidth communications,and standardized application management.Edge computing replaces traditional architectures quickly and become a new direction due to its advantages in connection,real-time control,intelligence,security.However,edge devices are limited in computing,network,storage and other aspects,which will lead to the inability to complete the computing tasks quickly.At the same time,the delay constraints of the cloud-edge communication process restrict the real-time information synchronization between cloud side and edge side,and it is difficult to achieve a high real-time resource optimization mechanism.We take the field of intelligent buildings as a specific application scenario,and researches cloud-edge collaborative management components from the aspects of overall architecture,edge device management,flexible resource allocation,and reliable data transmission.The main research contents are as follows:(a)Based on the investigation of edge computing and cloud-edge collaborative control strategy,the system architecture and key components of cloud-edge collaborative management for intelligent buildings are designed,which realize cloud information integration management and collaborative control for edge side subsystems.Meanwhile,aiming at the problem that the lighting,security and other subsystems in building control are difficult to manage in a unified manner,a standardized model for building management and control business is established.The characteristics of cloudedge transmission in intelligent building scenario are analyzed,which leads to an MQTT-based dual-broker communication architecture.(b)Faced with the problem of limited computing,network and storage resources of edge nodes,a task execution time minimization scheduling algorithm was designed,which is weakly dependent on cloud-edge information synchronization.It solves the problem that cloud side data transmission delay affects the real-time performance of task scheduling decision.The real-time task scheduling problem is converted to the task of minimizing execution time,and greedy-genetic algorithm is used to quickly solve the approximate global optimal scheduling decision.(c)A two-step coding scheme based on fountain code is designed to meet the realtime and transmission reliability requirements of large data packets such as container images and video files in the process of cloud-edge communication,which realizes stable and fast transmission of large data packets under unstable network environment.Two LT code(Luby Transform code)improvement methods based on excessive truncation optimization and degree distribution optimization are proposed,which optimize the decoding efficiency and transmission time.Combined with building control business scenarios,the key management components such as remote operation and maintenance,system configuration,task scheduling and data transmission under cloud-side collaboration are designed in this paper,which meet the application requirements of cloud-side collaboration control for intelligent buildings in function and performance tests.The average time of cloud instruction response is 25.1ms,while the average time of edge node instruction response is 7.2ms.By using the task scheduling strategy designed in this paper,the average execution time of computing tasks is reduced by 54.7%.The two-step coding scheme proposed in this paper can reduce the transmission time of big data by 40%.Compared with the classical LT code,the decoding efficiency is improved by 409% and the decoding time is reduced by 28%. |
