In this correspondence, a novel framework is proposed for analyzing data offloading in a multi-access edge computing system. Specifically, a two-phase algorithm, is proposed, including two key phases: 1) user association phase and 2) task offloading phase. In the first phase, a ruin theory-based approach is developed to obtain the users association considering the users' transmission reliability and resource utilization efficiency. Meanwhile, in the second phase, an optimization-based algorithm is used to optimize the data offloading process. In particular, ruin theory is used to manage the user association phase, and a ruin probability-based preference profile is considered to control the priority of proposing users. Here, ruin probability is derived by the surplus buffer space of each edge node at each time slot. Giving the association results, an optimization problem is formulated to optimize the amount of offloaded data aiming at minimizing the energy consumption of users. Simulation results show that the developed solutions guarantee system reliability, association efficiency under a tolerable value of surplus buffer size, and minimize the total energy consumption of all users.
翻译:在本文中,提出了一种分析多接入边缘计算系统中数据离载的新框架。具体而言,提出了一个由两个关键阶段组成的二阶段算法,包括: 1)用户关联阶段和2)任务离载阶段。在第一阶段中,开发了一种基于破产理论的方法,考虑用户的传输可靠性和资源利用效率来获取用户关联。同时,在第二阶段中,使用一个基于优化的算法来优化数据离载过程。特别地,利用破产理论来管理用户关联阶段,并考虑基于破产概率的偏好配置文件来控制提议用户的优先级。这里,破产概率是通过每个边缘节点在每个时隙的剩余缓冲空间来推导的。给出关联结果后,构建一个优化问题来优化离载的数据数量,以最小化所有用户的能量消耗。仿真结果表明,所开发的解决方案保证了系统可靠性,在容忍的剩余缓冲区大小下保证了关联效率,并最小化了所有用户的总能量消耗。