Decentralized AI: The role of edge intelligence in next-gen computing

Dhruvitkumar V. Talati *

Independent Researcher, USA.
 
Review
International Journal of Science and Research Archive, 2021, 02(01), 216-232.
Article DOI: 10.30574/ijsra.2021.2.1.0050
Publication history: 
Received on 10 February 2021; revised on 17 April 2021; accepted on 21 April 2021
 
Abstract: 
With the rapid development of communication technology, the explosive growth of mobile and IoT devices, and growing requirements for real-time data processing, a new paradigm of computing, Edge Computing, has appeared. It moves computing power in the direction of data sources to mitigate latency, bandwidth usage, and dependence on cloud computing. In parallel, Artificial Intelligence (AI) has progressed notably with deep learning technology, highly optimized hardware, and distributed computing paradigms to yield smart applications of high computational loads. Nonetheless, the huge amounts of data generated on the network edge impose heavy challenges in managing data, network optimization, and implementing AI models. This has pushed the convergence of Edge Computing and AI, which has led to a new research area called Edge Intelligence.
Edge Intelligence is further divided into two broad categories
·         AI for Edge (Intelligence-enabled Edge Computing) – This is concerned with augmenting Edge Computing architectures with AI-based methods, including resource management, task scheduling, computation offloading, and network optimization.
·         Edge AI (Artificial Intelligence on Edge) – This involves executing AI models on edge devices directly, enabling local training and inference with minimal dependence on the cloud, thereby improving privacy, efficiency, and real-time processing.
This paper provides an overview of Edge Intelligence, including fundamental concepts, future technologies, and research directions. We identify critical challenges such as efficient deployment of AI models, decentralized AI learning through federated learning, and edge-centric accelerations of domain-specific hardware, and discuss how Edge Intelligence has the potential to transform domains like autonomous systems, smart cities, factory automation, and wireless networks. This paper documents the present trend and future path to act as the basis for researchers, engineers, and industry players seeking to improve the topic of AI-driven Edge Computing.
 
Keywords: 
Edge Intelligence; Edge Computing; Artificial Intelligence; Wireless Networks; Distributed AI; Computation Offloading; Federated Learning; Real-Time AI; Model Optimization; Ai Acceleration
 
Full text article in PDF: 
 
Copyright information: 

Copyright © 2021 Author(s) retain the copyright of this article. This article is published under the terms of the Creative Commons Attribution Liscense 4.0