Jadwal Sholat

Memuat jadwal sholatโ€ฆ

Ilmu Komputer & AI editorial

Open AccessOA2022

Edge Computing with Artificial Intelligence: A Machine Learning Perspective

This ACM Computing Surveys paper explores how edge computing (EC) addresses the limitations of cloud computing for IoT data, and how artificial intelligence, particularly machine learning, can enhance EC performance and enable new applications.
H. Hua; Yu-Tong Li; Tonghe Wang; Nan-Qing Dong; Wei Li; Jun-Wei Caoยท ACM Computing Surveysยท 2022ยท DOI 10.1145/3555802

The core problem

The widespread adoption of the Internet of Things (IoT) has generated massive amounts of data, which has in turn fueled the development of artificial intelligence (AI). However, traditional cloud computing models struggle to handle the volume, latency, and bandwidth requirements of IoT data independently. Edge computing (EC) has emerged as a promising paradigm that brings computation closer to the data source, addressing these challenges. While early EC research relied on traditional (non-AI) methods, these approaches have shown limitations in optimizing EC performance. Recognizing the success of AI in various domains, EC researchers have increasingly turned to AI, especially machine learning (ML), to enhance EC. This article provides a comprehensive survey of the intersection between EC and AI, explaining the formal definition of EC, discussing key problems in EC, summarizing traditional solutions and their limitations, and highlighting how AI can be used to optimize EC and how AI applications can benefit from the EC architecture. The goal is to guide future research by exploiting the mutually beneficial relationship between AI and EC.

Innovation

The authors conduct a systematic literature review of edge computing and artificial intelligence, with a focus on machine learning. They first formalize the definition of EC and articulate the reasons for its emergence. Then, they identify core problems in EC, such as resource allocation, latency minimization, energy efficiency, and privacy preservation. For each problem, they summarize traditional solutions and analyze their limitations. Subsequently, they review research that applies AI/ML techniques to optimize EC, categorizing the approaches based on the type of learning (e.g., supervised, unsupervised, reinforcement learning) and the EC problem addressed. They also examine how EC can support AI applications, particularly in scenarios where low latency and data locality are critical. The survey methodology involves collecting and synthesizing relevant studies from academic databases, though the specific search strategy and inclusion criteria are not detailed in the provided abstract. The authors also discuss open challenges and future research directions.
Introduction
The widespread adoption of the Internet of Things (IoT) has generated massive amounts of data, which has in turn fueled the development of artificial intelligence (AI). However, traditional cloud computing models struggle to handle the volume, latency, and bandwidth requirements of IoT data independently. Edge computing (EC) has emerged as a promising paradigm that brings computation closer to the data source, addressing these challenges. While early EC research relied on traditional (non-AI) methods, these approaches have shown limitations in optimizing EC performance. Recognizing the success of AI in various domains, EC researchers have increasingly turned to AI, especially machine learning (ML), to enhance EC. This article provides a comprehensive survey of the intersection between EC and AI, explaining the formal definition of EC, discussing key problems in EC, summarizing traditional solutions and their limitations, and highlighting how AI can be used to optimize EC and how AI applications can benefit from the EC architecture. The goal is to guide future research by exploiting the mutually beneficial relationship between AI and EC.

Why it matters

The authors discuss the mutually beneficial relationship between EC and AI: AI can optimize EC performance, and EC can support AI applications. They emphasize that while AI brings powerful tools, it also introduces challenges such as model interpretability, energy consumption, and security. For instance, deploying complex ML models on resource-constrained edge devices requires model compression and efficient inference techniques. The survey also identifies open research directions, including the need for lightweight AI models, robust and secure federated learning, and the integration of EC with 5G/6G networks. The authors argue that the synergy between EC and AI will drive the development of intelligent edge systems, enabling applications like autonomous vehicles, smart cities, and industrial IoT. They conclude that interdisciplinary research combining AI, networking, and distributed systems is essential to fully realize the potential of edge intelligence.

Who should read this

Pembaca Research Digest dan praktisi rumpun ini.

Opening member contentโ€ฆ