Joint Optimization of DNN Inference Delay and Energy under Accuracy Constraints for AR Applications

Guangjin Pan; Heng Zhang; Shugong Xu; Shunqing Zhang; Xiaojing Chen

doi:10.1109/GLOBECOM48099.2022.10001451

Joint Optimization of DNN Inference Delay and Energy under Accuracy Constraints for AR Applications

Guangjin Pan
, Heng Zhang
, Shugong Xu^*
, Shunqing Zhang
, Xiaojing Chen

^*Corresponding author for this work

Shanghai University

Research output: Contribution to journal › Conference article › peer-review

6 Citations (Scopus)

Abstract

The high computational complexity and high energy consumption of artificial intelligence (AI) algorithms hinder their application in augmented reality (AR) systems. This paper considers the scene of completing video-based AI inference tasks in the mobile edge computing (MEC) system. We use multiply-and-accumulate operations (MACs) for problem analysis and optimize delay and energy consumption under accuracy constraints. To solve this problem, we first assume that offloading policy is known and decouple the problem into two subproblems. After solving these two subproblems, we propose an iterative-based scheduling algorithm to obtain the optimal offloading policy. We also experimentally discuss the relationship between delay, energy consumption, and inference accuracy.

Original language	English
Pages (from-to)	2230-2235
Number of pages	6
Journal	Proceedings - IEEE Global Communications Conference, GLOBECOM
DOIs	https://doi.org/10.1109/GLOBECOM48099.2022.10001451
Publication status	Published - 2022
Externally published	Yes
Event	2022 IEEE Global Communications Conference, GLOBECOM 2022 - Rio de Janeiro, Brazil Duration: 4 Dec 2022 → 8 Dec 2022

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 7 Affordable and Clean Energy

Keywords

Mobile augmented reality
edge intelligence
mobile edge computing
resource allocation

Access to Document

10.1109/GLOBECOM48099.2022.10001451

Cite this

@article{047ae67fd0224768b762a745b950949d,

title = "Joint Optimization of DNN Inference Delay and Energy under Accuracy Constraints for AR Applications",

abstract = "The high computational complexity and high energy consumption of artificial intelligence (AI) algorithms hinder their application in augmented reality (AR) systems. This paper considers the scene of completing video-based AI inference tasks in the mobile edge computing (MEC) system. We use multiply-and-accumulate operations (MACs) for problem analysis and optimize delay and energy consumption under accuracy constraints. To solve this problem, we first assume that offloading policy is known and decouple the problem into two subproblems. After solving these two subproblems, we propose an iterative-based scheduling algorithm to obtain the optimal offloading policy. We also experimentally discuss the relationship between delay, energy consumption, and inference accuracy.",

keywords = "Mobile augmented reality, edge intelligence, mobile edge computing, resource allocation",

author = "Guangjin Pan and Heng Zhang and Shugong Xu and Shunqing Zhang and Xiaojing Chen",

note = "Publisher Copyright: {\textcopyright} 2022 IEEE.; 2022 IEEE Global Communications Conference, GLOBECOM 2022 ; Conference date: 04-12-2022 Through 08-12-2022",

year = "2022",

doi = "10.1109/GLOBECOM48099.2022.10001451",

language = "English",

pages = "2230--2235",

journal = "Proceedings - IEEE Global Communications Conference, GLOBECOM",

issn = "2334-0983",

}

TY - JOUR

T1 - Joint Optimization of DNN Inference Delay and Energy under Accuracy Constraints for AR Applications

AU - Pan, Guangjin

AU - Zhang, Heng

AU - Xu, Shugong

AU - Zhang, Shunqing

AU - Chen, Xiaojing

PY - 2022

Y1 - 2022

N2 - The high computational complexity and high energy consumption of artificial intelligence (AI) algorithms hinder their application in augmented reality (AR) systems. This paper considers the scene of completing video-based AI inference tasks in the mobile edge computing (MEC) system. We use multiply-and-accumulate operations (MACs) for problem analysis and optimize delay and energy consumption under accuracy constraints. To solve this problem, we first assume that offloading policy is known and decouple the problem into two subproblems. After solving these two subproblems, we propose an iterative-based scheduling algorithm to obtain the optimal offloading policy. We also experimentally discuss the relationship between delay, energy consumption, and inference accuracy.

AB - The high computational complexity and high energy consumption of artificial intelligence (AI) algorithms hinder their application in augmented reality (AR) systems. This paper considers the scene of completing video-based AI inference tasks in the mobile edge computing (MEC) system. We use multiply-and-accumulate operations (MACs) for problem analysis and optimize delay and energy consumption under accuracy constraints. To solve this problem, we first assume that offloading policy is known and decouple the problem into two subproblems. After solving these two subproblems, we propose an iterative-based scheduling algorithm to obtain the optimal offloading policy. We also experimentally discuss the relationship between delay, energy consumption, and inference accuracy.

KW - Mobile augmented reality

KW - edge intelligence

KW - mobile edge computing

KW - resource allocation

UR - https://www.scopus.com/pages/publications/85146927353

U2 - 10.1109/GLOBECOM48099.2022.10001451

DO - 10.1109/GLOBECOM48099.2022.10001451

M3 - Conference article

AN - SCOPUS:85146927353

SN - 2334-0983

SP - 2230

EP - 2235

JO - Proceedings - IEEE Global Communications Conference, GLOBECOM

JF - Proceedings - IEEE Global Communications Conference, GLOBECOM

T2 - 2022 IEEE Global Communications Conference, GLOBECOM 2022

Y2 - 4 December 2022 through 8 December 2022

ER -

Joint Optimization of DNN Inference Delay and Energy under Accuracy Constraints for AR Applications

Abstract

UN SDGs

Keywords

Access to Document

Other files and links

Fingerprint

Cite this