Joint Cache Content Placement and Task Offloading in C-RAN Enabled by Multi-Layer MEC

Haibo Mei, Kezhi Wang, Kun Yang

Research output: Contribution to journalArticlepeer-review

14 Citations (Scopus)
18 Downloads (Pure)

Abstract

In this paper, we work on a Cache and Multi-layer MEC enabled C-RAN (CMM-CRAN) to handle various user tasks with minimized latency and energy cost. We intend to solve two particular problems of CMM-CRAN. First, because CMM-CRAN has to maximally cache the most frequently requested data from Service Provide Server (SPS) to Remote Radio Head (RRH) and later offered to proximity mobile users, the cache content placement from SPSs to RRHs becomes a many-to-many matching problem with peer effects. Second, because of multi-layer MEC, a user task has to be dynamically controlled to be offloaded to the best fit cloud, i.e., either local MEC or remote MEC, to get served. This dynamic task offloading is a Multi-Dimension Multiple-Choice Knapsack (MMCK) problem. To solve these two problems, we provide a Joint Cache content placement and task Offloading Solution (JCOS) to CMM-CRAN that utilizes Proportional Fairness (PF) as the user scheduling policy. JCOS applies a Gale-Shaply (GS) method to work out the cache content placement, and a Population Evolution (PE) game theory coupled with a use of Analytic Hierarchy Process(AHP) to work out the dynamic user task offloading. According to the simulation results, CMM-CRAN with JCOS is proved to be able to provide highly desired low-latency communication and computation services with decreased energy cost to mobile users.
Original languageEnglish
Article number1826
JournalSensors
Volume18
Issue number6
DOIs
Publication statusPublished - 5 Jun 2018

Keywords

  • cache content placement
  • user task offloading
  • Gale-Shaply method
  • population evolution game theory

Fingerprint

Dive into the research topics of 'Joint Cache Content Placement and Task Offloading in C-RAN Enabled by Multi-Layer MEC'. Together they form a unique fingerprint.

Cite this