去蜂窝大规模MIMO辅助的移动边缘计算系统计算任务卸载与分配策略

Computation offloading and allocation strategy for cell-free massiveMIMO-enabled mobile edge computing systems

面向B5G和6G的新兴网络架构和技术服务需求,将去蜂窝大规模多输入多输出(cell-free massive MIMO,CF-mMIMO)赋能于移动边缘计算(mobile edge computing,MEC),有助于处理分布式物联网中的计算密集型和延迟敏感型任务。针对CF-mMIMO辅助的MEC系统,在能量限制下意在最大限度地减少完成不同任务类型的计算任务的延迟。为完成以上目标,设计了一种基于本地设备(user equipment,UE)、多接入点(access point,AP)和中心处理器的云-边-端协作的任务卸载策略。具体地,首先根据每个UE和AP服务的不同数据类型,利用凸优化和图匹配方法交替迭代,进行卸载关联和任务比例的优化;然后在回传链路的限制下,提出一种改进的二进制鲸鱼优化算法,将未分配终端和关联接入点任务进一步卸载至处理高效的云端。所提算法相较于蚁群优化算法、混合灰狼优化算法等其他的元启发式效果更优,在离散的卸载优化问题上表现较好,可以为分布式网络提供良好的卸载优化策略并大幅度降低整体网络的平均时延。

Emerging network architectures and technical service requirements for B5G and 6G will enable MEC with CF-mMIMO,helping to handle compute-intensive and latency-sensitive tasks in distributed IoT.For CF-mMIMO-assisted MEC systems,this paper aimed to minimize the delay in completing computational tasks of different task types under energy constraint.In order to solve the above goals,this paper designed a task offloading strategy based on UEs,multiple APs and CPU(central processing unit)for cloud-edge-end collaboration.Specifically,according to the different data types of each UE and AP service,this paper firstly used the convex optimization and graph matching methods to alternately iterate to optimize the offload association and task ratio.Then,under the limitation of the backhaul link,this paper used an improved binary whale optimization algorithm to further offload the tasks of unallocated terminals and associated access points to the cloud with efficient processing.Compared with other meta-heuristics such as ant colony optimization algorithm and hybrid gray wolf optimization algorithm,the proposed algorithm has better performance on discrete offload optimization problems,which can provide a good offload optimization strategy for distributed systems and greatly reduce the average delay of the whole network.