期刊
IEEE TRANSACTIONS ON NETWORK AND SERVICE MANAGEMENT
卷 18, 期 2, 页码 2089-2106出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TNSM.2021.3072433
关键词
Resource management; Delays; Containers; Real-time systems; Task analysis; Computational modeling; Servers; Mobile networks; multi-access edge computing; offloading; mobility management; resource allocation; prediction; real-time
资金
- Ministry of Education, Youth and Sports, Czech Republic [LTT18007]
- [SGS17/184/OHK3/3T/13]
- [SGS20/169/OHK3/3T/13]
This paper proposes a low-complexity computing and communication resource allocation method for real-time computing tasks generated by mobile users, utilizing probabilistic modeling of user movement to achieve resource pre-allocation and selection of suitable communication paths, keeping the edge computing delay below 100 ms and enhancing support for real-time applications.
The Multi-Access Edge Computing (MEC) constitutes computing over virtualized resources distributed at the edge of mobile network. For mobile users, an optimal allocation of communication and computing resources changes over time and space, and the resource allocation becomes a complex problem. Moreover, for delay constrained applications, the resource allocation to mobile users cannot be solved by approaches designed for static users, as a solution would not be obtained within a desired time. Thus, in this paper, we propose a low-complexity computing and communication resource allocation for offloading of real-time computing tasks generated with a high arrival rate by the mobile users. We exploit probabilistic modeling of the users' movement to pre-allocate the computing resources at base stations and to select suitable communication paths between the users and the base station with the pre-allocated computing resources. The simulations show that the proposed algorithm keeps the offloading delay below 100 ms for the small tasks even with the arrival rate of five tasks per second per user, while the state-of-the-art algorithms can handle only up to 0.5 tasks per second per user. Thus, the proposal enables an exploitation of the MEC for various real-time applications even if the users are moving.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据