As one of the key communication scenarios, ultrareliable low-latency communication (uRLLC) has become an important pillar to promote the vigorous development of intelligent mobile communications. In the practical scenarios, uRLLC services have strict and diverse Quality-of-Service (QoS) requirements. However, the existing networks are difficult to meet the various delay and reliability requirements of uRLLC services. Moreover, the improvement of performance should not ignore the shortage of resources. A flexible and on-demand network solution is quite necessary, which could provide customized services according to the specific requirements and maximize the utilization efficiency of network resources. In this article, we propose an intelligent and flexible network solution (IFNS) based on the stochastic network calculus (SNC) model. Three key technologies are considered in the IFNS, that are flexible transmission time interval scheduling, flexible packet duplication transmission, and rate-adaptive reliable transmission. While providing customized services for users with various requirements, it realizes the balance between system energy efficiency and spectral efficiency and improves the resource utilization efficiency of the network. Based on the basic domain knowledge and the past experience, we propose the knowledge-assistance meta actor-critic (K-MAC) algorithm to solve the complex optimization problem caused by SNC modeling. Finally, simulation results show that the performance of the IFNS is improved 23.15%, the K-MAC algorithm has good convergence performance and reduces the complexity up to 89.4% compared with common learning algorithms.
All Science Journal Classification (ASJC) codes