Optimize the placement of edge server between workload balancing and system delay in smart city

• Published in: Peer-to-peer networking and applications Vol. 14; no. 6; pp. 3778 - 3792
• Main Authors: Zhao, Xingbing, Zeng, Yu, Ding, Hongwei, Li, Bo, Yang, Zhijun
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.11.2021; Springer Nature B.V
• Subjects: Communication; Communication networks; Communications Engineering; Communications networks; Computer Communication Networks; Data processing; Data storage; Data transmission; Delay; Edge computing; Engineering; Genetic algorithms; Information Systems and Communication Service; Internet of Things; Load balancing; Multiple objective analysis; Networks; Optimization; Placement; Radio equipment; Servers; Signal,Image and Speech Processing; Sorting algorithms; Special Issue on Convergence of Edge Computing and Next Generation Networking; Workload; Workloads; NSGA-II; Edge server; Placement; Mobile edge computing; Edge computing
• ISSN: 1936-6442; 1936-6450
• DOI: 10.1007/s12083-021-01208-0

With the advent of mobile Internet and IoT era, various smart terminals generate a large amount of data at the edge of the network, and how to transmit and process these data at high speed poses a challenge to the traditional communication networks. Edge computing, as an emerging framework, can improve the communication capability and data processing capacity of traditional communication networks by improving their architecture. Edge server placement (ESP) technology is one of the key technologies of edge computing, which can effectively reduce data transmission delay and improve data processing efficiency by placing edge servers (ESs) with computing and data storage functions at base stations to sink some functions of the core network to the edge of the network. In this paper, we study the k edge servers placement problem (KESP problem) in smart cities. We first elaborate it as a multi-objective optimization problem for optimal workload balancing and system delay under constraints. Then a modified multi-objective non-dominated sorting genetic algorithm with elite policy (MNSGA-II) is proposed to optimize this problem. Finally, simulations are performed based on real network datasets. The simulation results show that MNSGA-II reduces the system overhead by about 38.4%, 40.6%, and 59.3% on average compared to Random, K-Means, and Top-K.