| Due to the rapid development of information technology,some emerging services with higher requirements for delay and reliability will inevitably appear.As a new generation of mobile communication technology,5G will be widely used in scenarios such as enhanced mobile bandwidth(eMBB),ultra-high reliability and low-latency communication(URLLC),and massive Io T(m MTC).According to the 5G development strategy,at the initial stage of deployment,the coexistence of eMBB and URLLC services will become a typical application scenario.Among them,the eMBB service is an upgrade of the traditional mobile network,which is characterized by a higher data transmission rate and resource occupancy rate;the URLLC service is a new application scenario proposed with the development of 5G.Compared with the previous mobile communication,its delay and reliability all have been greatly improved and can be widely used in vertical industries such as the Internet of Vehicles,smart grid and industrial automation.Since both eMBB and URLLC services have wireless transmission and task computing requirements,the limited frequency spectrum and computing resources in the communication system will face fierce competition.Therefore,while meeting the different service quality requirements of the above two businesses,reasonable resource allocation for the two businesses is the key issue studied in this article.This thesis starts research on the basis of existing resource allocation technology.The main research content and innovation results include:Aiming at the problem of impaired eMBB transmission rate caused by resource competition in the coexistence of eMBB and URLLC services,a resource allocation algorithm based on the Penalty Concave Convex Process(PCCP)is designed.First,the resource allocation problem is described as an optimization problem with constraints,the optimization problem is approximately transformed to obtain an equivalent problem that meets the PCCP algorithm framework,and the optimization problem is solved according to the iterative algorithm based on PCCP.Secondly,in order to overcome the complexity of the proposed optimization problem,a low-complexity heuristic scheduling algorithm is proposed.Finally,the performance of the proposed algorithms is evaluated and compared through simulation to show that they can dynamically allocate resources for different services while meeting the requirements of reliability,delay and minimum speed.Aiming at the problem of eMBB and URLLC business task offloading and resource joint allocation,a joint optimization algorithm based on deep reinforcement learning is designed.First,by considering task offloading,user delay,and resource allocation,a joint optimization problem with the goal of minimizing system cost is formed.The calculation cost is defined by the weighted sum of delay and energy consumption.Secondly,in order to simplify the solution of the optimization problem,the proposed scheme divides the task offloading and resource allocation problem into two stages to solve the problem.The first stage uses an improved algorithm based on Deep Q Network(DQN)to obtain the user’s task offloading decision,and the second stage uses Greedy strategy and Lagrangian multiplier method for joint resource allocation.Finally,through simulation verification,the proposed scheme also has advantages in reducing the computational cost of the system under the premise of meeting the delay requirements of each user. |
