Joint Resource, Deployment, and Caching Optimization for AR Applications in Dynamic UAV NOMA Networks

The cache-enabling unmanned aerial vehicle (UAV) non-orthogonal multiple access (NOMA) networks for mixture of augmented reality (AR) and normal multimedia applications are investigated, which is assisted by UAV base stations. The user association, power allocation of NOMA, deployment of UAVs and ca...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on wireless communications 2022-05, Vol.21 (5), p.3409-3422
Main Authors: Zhang, Tiankui, Wang, Ziduan, Liu, Yuanwei, Xu, Wenjun, Nallanathan, Arumugam
Format: Article
Language:English
Subjects:
Algorithms
Augmented reality
Caching
Deep deterministic policy gradient
Delay
Delays
edge caching
Game theory
Heuristic algorithms
Machine learning
Multimedia
NOMA
non-orthogonal multiple access
Nonorthogonal multiple access
Optimization
Placement
Resource management
Stackelberg game
unmanned aerial vehicle
Unmanned aerial vehicles
Wireless communication
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The cache-enabling unmanned aerial vehicle (UAV) non-orthogonal multiple access (NOMA) networks for mixture of augmented reality (AR) and normal multimedia applications are investigated, which is assisted by UAV base stations. The user association, power allocation of NOMA, deployment of UAVs and caching placement of UAVs are jointly optimized to minimize the content delivery delay. A branch and bound (BaB) based algorithm is proposed to obtain the per-slot optimization. To cope with the dynamic content requests and mobility of users in practical scenarios, the original optimization problem is transformed to a Stackelberg game. Specifically, the game is decomposed into a leader level user association sub-problem and a number of power allocation, UAV deployment and caching placement follower level sub-problems. The long-term minimization was further solved by a deep reinforcement learning (DRL) based algorithm. Simulation result shows that the content delivery delay of the proposed BaB based algorithm is much lower than benchmark algorithms, as the optimal solution in each time slot is achieved. Meanwhile, the proposed DRL based algorithm achieves a relatively low long-term content delivery delay in the dynamic environment with lower computation complexity than BaB based algorithm.
ISSN:1536-1276
1558-2248
DOI:10.1109/TWC.2021.3121584