Optimization of Load-Balancing Strategy by Self-Powered Sensor and Digital Twins in Software-Defined Networks

This work is aimed at smart sensors in the context of the boom in intelligent manufacturing under artificial intelligence (AI). First, an intelligent industrial system is innovatively constructed based on AI and digital twins (DTs) technology. Its sensors use self-powered sensors based on solar energy. Second, layered architecture and four modules of the balancing system are built based on the software-defined network (SDN) with programmability and flexibility, and a controller load-balancing strategy is proposed based on the layered architecture. Besides, the link load-balancing algorithm is optimized based on Dijkstra’s algorithm (this algorithm is the leading AI algorithm used here). Finally, simulation experiments are carried out on different controllers and algorithms. The results show that the AI system has a high control accuracy. Controller 5 experiences a surge in load at 45 s, exceeding the load threshold. Controllers 1 and 5 are selected for load balancing. Migration can be done quickly when load balancing is completed within 5 s. Control transfer load balance (CTLB) has the longest response time when the load value reaches 20%–70%. The load policy has the shortest response time, indicating the best effect. In addition, the load-balancing algorithm proposed here fully considers the link conditions in the link load comparison and achieves an excellent effect. The load rate increase affects the network delay’s continuous increase. Moreover, the highest network delay of the algorithm reported here is 0.25 ms, that of the Yen algorithm is 0.31 ms, and that of Dijkstra’s algorithm is 0.43 ms, which is 83.3% lower than that of the unoptimized algorithm. In the 10%–100% load rate range, the packet loss rate of the proposed algorithm is lower than that of the other two algorithms. Load-balancing strategies outperform other standard algorithms in terms of network performance. This study provides some references for the development of self-powered sensors in the field of industrial manufacturing.