An MA-PPO based Task Scheduling Method for Computing First Networks
Jialu Zhao, Kun Xie, Huiyue Sun, Chengcheng Li, Yasheng Zhang · 2022
Aiming at converging computing and network resources, computing first networking (CFN) provides the computing task scheduling solution along with routing policy based on real-time computing and network information, to achieve the optimal utilization of overall resources and improve user experience. Most of the existing schemes consider computing and network resources separately to obtain their optimal results, lacking a load balancing mechanism that comprehensively considers both. In view of the above, this paper proposes an MA-PPO-based service scheduling algorithm for CFN. Based on the current computing and network resource status, scheduling strategies combined with computing task allocation and routing planning are flexibly generated for each computing task to achieve the overall optimization of computing and network resources utilization. The experimental results show that the algorithm can effectively improve the utilization of computing and network resources, shorten task processing time, and improve user experience.