Content-Aware User Association and Multi-User MIMO Beamforming over Mobile Edge Caching

Mobile edge caching (MEC) has been introduced to support ever-growing end-users' needs. To reduce the backhaul traffic demand and content delivery latency, cache-enabled edge servers at base stations (BSs) are employed to provision popular contents at the network edge. In this paper, multiple-i...

Full description

Saved in:
Bibliographic Details
Published inarXiv.org
Main Authors Mosleh, Susanna, Fan, Qiang, Liu, Lingjia, Ashdown, Jonathan D, Perrins, Erik, Turck, Kurt
Format Paper
LanguageEnglish
Published Ithaca Cornell University Library, arXiv.org 30.12.2020
Subjects
Caching
Computer simulation
Iterative algorithms
Mixed integer
Optimization
Placement
Online AccessGet full text

Cover

More Information
Summary:Mobile edge caching (MEC) has been introduced to support ever-growing end-users' needs. To reduce the backhaul traffic demand and content delivery latency, cache-enabled edge servers at base stations (BSs) are employed to provision popular contents at the network edge. In this paper, multiple-input-multiple-output (MIMO) operation and user association policy are linked to the underlying cache placement strategy to ensure a good trade-off between load balancing and backhaul traffic taking into account the underlying wireless channel and the finite cache capacity at edge servers. Due to the coupled interference among mobile stations, the binary nature of the underlying cache placement and user association matrices, the resulting mixed-timescale mixed integer optimization problem is nonconvex and NP-hard. To solve this problem, we decompose the joint optimization problem into a long-term content placement sub-problem and a short-term content delivery sub-problem. A novel iterative algorithm is introduced by leveraging the alternating direction method of multipliers together with a stochastic parallel successive convex approximation-based algorithm. The introduced scheme enables all BSs to update their optimization variables in parallel by solving a sequence of convex subproblems. Simulation evaluation demonstrates the efficiency of our strategy.
ISSN:2331-8422