ISSN (Online) 2581-9429

IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)
International Open-Access, Double-Blind, Peer-Reviewed, Refereed, Multidisciplinary Online Journal
Impact Factor: 7.301 Volume 3, Issue 1, September 2023

AI-Driven Edge Computing for IoT: A

Comprehensive Survey and Future Directions
Shalini Saha1, Kushal Banerjee2, Sandhita Ghosh3, Samriddha Mitra4, Debrupa Pal5
Student, Computer Application, Narula Institute of Technology, Kolkata, India1234
Assistant Professor, Computer Application, Narula Institute of Technology, Kolkata, India5

Abstract: The proliferation of Internet of Things (IoT) devices has ushered in an era of unprecedented data
generation and processing demands. To meet these challenges, the convergence of Artificial Intelligence
(AI) and Edge Computing has emerged as a promising solution. This paper presents a comprehensive
survey of the state-of-the-art in AI-Driven Edge Computing for IoT, exploring key technologies,
architectures, and applications. We begin by providing an overview of the fundamental concepts
underpinning IoT and edge computing, highlighting the convergence of these technologies. As IoT
continues to expand its reach across diverse domains, understanding the synergy between AI and edge
computing is paramount for unlocking the full spectrum of possibilities, driving innovation, and ensuring a
more connected and intelligent future

Keywords: AI-driven edge computing, Internet of Things (IoT), Edge devices, Machine learning, Deep
learning

I. INTRODUCTION
The Internet of Things (IoT) has ushered in an era of unprecedented connectivity, transforming our world by
interconnecting billions of devices, sensors, and systems. This interconnected ecosystem generates vast amounts of
data, promising insights and efficiencies across numerous domains, from healthcare and agriculture to transportation
and smart cities. However, this deluge of data presents a significant challenge: how to process, analyse, and extract
meaningful insights from this data in real-time or near real-time.
Enter edge computing, a paradigm that pushes computational capabilities closer to the data source, reducing latency,
bandwidth usage, and enhancing privacy and security. While edge computing has alleviated some of the challenges
posed by IoT, it is artificial intelligence (AI) that has emerged as the transformative force, enabling intelligent decision-
making at the edge. AI-Driven Edge Computing is poised to unlock the full potential of IoT by equipping devices with
advanced analytics, machine learning algorithms, and autonomous decision-making capabilities. This comprehensive
survey paper aims to provide a thorough exploration of the synergistic relationship between AI and edge computing in
the context of IoT.

II. IOT EDGE DEVICES AND INFRASTRUCTURE

IoT (Internet of Things) edge devices and infrastructure play a crucial role in the rapidly evolving landscape of
connected technologies. These components form the foundation for collecting, processing, and transmitting data at the
edge of a network, closer to the data source, which offers various advantages.
Edge devices are the physical endpoints in an IoT ecosystem, such as sensors, actuators, cameras, and other embedded
systems. These devices gather real-time data from the physical world, making it available for analysis and decision-
making. Edge devices are typically equipped with microcontrollers or small processors that enable data processing and
minimal local decision-making, reducing the need to transmit all data to centralized cloud servers.
The infrastructure supporting IoT edge devices includes edge computing resources and communication protocols. Edge
computing involves deploying compute and storage resources closer to the edge devices. This allows for faster data
processing, reduced latency, and improved privacy and security by keeping sensitive data local. Popular edge
computing platforms, such as Azure IoT Edge and AWS Greengrass, provide a framework for deploying and managing
edge applications.

Copyright to IJARSCT DOI: 10.48175/IJARSCT-12921 117

www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)
International Open-Access, Double-Blind, Peer-Reviewed, Refereed, Multidisciplinary Online Journal
Impact Factor: 7.301 Volume 3, Issue 1, September 2023

Figure 1: Diagram of IoT Edge Architecture

Communication protocols are essential for connecting and transferring data between edge devices and centralized
systems. Common protocols include MQTT, CoAP, and HTTP, which facilitate efficient data exchange while
considering the constraints of edge devices, like limited bandwidth and intermittent connectivity. [1-2]
In conclusion, IoT edge devices and infrastructure are critical components in the IoT ecosystem, enabling real-time data
processing, reducing latency, and enhancing overall system reliability. As the IoT continues to grow, the development
and optimization of edge devices and infrastructure will be pivotal in shaping the future of connected technologies.

III. AI TECHNIQUES FOR IOT EDGE COMPUTING

AI techniques for IoT edge computing play a crucial role in enhancing the capabilities of edge devices, allowing them
to process and analyse data locally without relying solely on cloud resources. Here are four key points:
1. Edge Device Optimization: AI techniques are used to optimize edge devices' performance and resource
utilization. This includes deploying lightweight machine learning models that are designed to run efficiently
on constrained hardware, such as microcontrollers and low-power processors. Techniques like quantization
and model pruning reduce the computational and memory requirements of AI models, making them suitable
for edge devices with limited resources.
2. Real-time Data Processing: Edge computing relies on real-time data analysis to make immediate decisions. AI
algorithms like deep learning and neural networks are employed to process sensor data, images, and videos
locally. This enables quick responses to critical events without the latency associated with sending data to the
cloud and waiting for a response. For example, in smart cities, edge devices can use AI to analyse traffic
patterns and adjust traffic signals in real-time for optimal traffic flow. [3]
3. Anomaly Detection and Predictive Maintenance: AI-driven anomaly detection models are employed at the
edge to identify deviations from normal behaviour in IoT data streams. By continuously monitoring sensor
data, these models can detect potential issues or security threats early, preventing system failures or data
breaches. In industrial settings, predictive maintenance models analyse equipment sensor data to predict when
machines will require maintenance, reducing downtime and maintenance costs.
4. Edge-to-Cloud Collaboration: AI techniques enable a collaborative approach between edge devices and cloud
resources. Edge devices preprocess and filter data locally using AI algorithms before sending relevant
information to the cloud for further analysis and storage. This reduces the amount of data transmitted to the
cloud, saving bandwidth and lowering cloud computing costs. Additionally, AI models deployed in the cloud
can provide insights and predictions based on aggregated data from multiple edge devices, enhancing overall
system intelligence.
In summary, AI techniques for IoT edge computing empower edge devices with the capability to process data
efficiently, make real-time decisions, detect anomalies, and collaborate with cloud resources. These techniques are
crucial for harnessing the full potential of edge computing in various applications, ranging from smart cities and
industrial automation to healthcare and autonomous vehicles. [4]
Copyright to IJARSCT DOI: 10.48175/IJARSCT-12921 118
www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)
International Open-Access,
Access, Double-Blind,
Double Peer-Reviewed,
Reviewed, Refereed, Multidisciplinary Online Journal
Impact Factor: 7.301 Volume 3, Issue 1, September 2023

IV. AI-DRIVEN
DRIVEN EDGE COMPUTING
COMPUTING APPLICATIONS IN IO IOT
AI-driven
driven edge computing is revolutionizing the Internet of Things (IoT) landscape by bringing intelligenc
intelligence closer to
the data source, enabling real-time
time processing and decision making. Here are six key applications of AI
decision-making. AI-driven edge
computing in IoT:
1. Real-time Analytics: Edge devices equipped with AI can process data locally, allowing for immediate insight insights
and actions without the latency of sending data to the cloud. This is crucial in applications like predictive
maintenance in manufacturing.
2. Enhanced Security: AI at the edge can analyse video feeds, sensor data, and audio inputs to detect anomalies
time, bolstering the security of smart cities and buildings.
and security threats in real-time,
3. Energy Efficiency: AI-powered
powered edge devices can optimize energy consumption by adjusting settings in real real-
time based on occupancy, weather, and usage patterns, resulting
resulting in significant energy savings in smart homes
and offices.

Figure 2:
2 Diagram of AI-Driven Edge Computing Applications
5. Autonomous Vehicles: Edge AI plays a pivotal role in self-driving
self driving cars by processing data from cameras,
LiDAR, and sensors on-board,
board, allowing the vehicle to make instant decisions without relying solely on cloud
cloud-
based processing.
6. Healthcare Monitoring: Wearable devices with with AI at the edge can monitor vital signs and provide immediate
feedback to patients or alert healthcare providers in case of emergencies, improving patient care and reducing
hospital admissions.
7. Agriculture: In precision agriculture, AI-powered
AI edge devices
ices analyse data from soil sensors, drones, and
cameras to optimize irrigation, fertilization, and pest control, leading to increased crop yields and resource
efficiency.
driven edge computing in IoT not only reduces data transmission costs but also en
AI-driven enhances
hances responsiveness, reliability,
changer in various industries. [5-6]
and privacy, making it a game-changer

V. CHALLENGES IN AI-DRIVEN
AI EDGE COMPUTING G FOR IOT
AI-driven
driven edge computing for IoT faces several challenges. Firstly, the limited computational resources available at the
edge devices restrict the complexity of AI models that can be deployed, impacting their accuracy and performance.
Secondly, ensuring data privacy and security is crucial, as sensitive information is processed closer to the sourcesource,
making it more vulnerable to potential breaches. Interoperability is another hurdle, as diverse IoT devices and platforms
must seamlessly integrate with AI solutions. Additionally, managing the scalability and reliability of edge computing
systems is challenging,
llenging, especially when dealing with a vast number of IoT devices.Power efficiency is a concern, as
energy-constrained
constrained edge devices must balance processing demands with power consumption. Latency can be
r time AI analysis. Finally, regulatory and ethical considerations
problematic, especially in applications requiring real-time
must be addressed to ensure AI-driven
driven edge computing adheres to legal frameworks and ethical standards, particularly
Copyright to IJARSCT DOI: 10.48175/IJARSCT-12921 119
www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)
International Open-Access, Double-Blind, Peer-Reviewed, Refereed, Multidisciplinary Online Journal
Impact Factor: 7.301 Volume 3, Issue 1, September 2023

when it comes to data collection and usage. These challenges highlight the need for ongoing research and development
in AI-driven edge computing to unlock its full potential in the IoT landscape. [7]

VI. CONCLUSION
In conclusion, this comprehensive survey on AI-driven edge computing for IoT has provided valuable insights into the
current state of this dynamic field and has shed light on its promising future directions. Through an in-depth
examination of the existing literature and a thorough analysis of key trends and challenges, we have explored the
synergy between AI and edge computing in addressing the unique requirements of IoT applications.
Throughout this survey, we have witnessed the tremendous potential of AI-driven edge computing to enhance the
efficiency, reliability, and intelligence of IoT systems. By pushing data processing and decision-making closer to the
data source, edge computing not only reduces latency and bandwidth consumption but also enables real-time analytics,
which is crucial for time-sensitive applications such as autonomous vehicles, smart cities, and industrial automation.
Moreover, AI algorithms and techniques, ranging from machine learning to deep learning, have shown remarkable
promise in enabling predictive maintenance, anomaly detection, and adaptive resource allocation within edge devices.
These capabilities empower IoT systems to operate more autonomously, adapt to changing conditions, and make data-
driven decisions, ultimately improving their overall performance and user experience.
However, this survey has also highlighted several challenges that need to be addressed in the future development of AI-
driven edge computing for IoT. These challenges include security and privacy concerns, resource constraints on edge
devices, interoperability issues, and the need for standardized architectures and protocols. Overcoming these hurdles
will be essential to fully realize the potential of this technology.
Looking ahead, the future of AI-driven edge computing for IoT is undeniably bright. As we continue to advance in AI
research and edge computing technologies, we can expect to see even more sophisticated and efficient solutions
emerge. Moreover, the integration of 5G networks and the proliferation of edge computing infrastructure will further
accelerate the adoption of AI in IoT. [8-10]

VII. ACKNOWLEDGMENT
We would like to express our sincere gratitude to all those who supported and contributed to the completion of this
research project. Without their invaluable assistance, guidance, and encouragement, this work would not have been
possible.
First and foremost, we extend our deepest appreciation to Srinjoy Saha, Student of Narula Institute of Technology, for
her continuous mentorship, insightful feedback, and unwavering support throughout the duration of this research. Her
expertise and guidance played a pivotal role in shaping the direction of this study. Apart from that, this research work is
not funded by any organization or any person.
We are also thankful to the members of the computer application department for their valuable discussions, critical
insights, and collaboration that enriched our understanding and perspective on the subject matter.Furthermore, we
would like to thank our friends and family for their unwavering encouragement and understanding during the often-
demanding phases of this project. Their emotional support provided the necessary foundation for our perseverance.

REFERENCES
[1]. M. Satyanarayanan, "The Emergence of Edge Computing," Computer, vol. 50, no. 1, pp. 30-39, 2017.
[2]. Y. Mao, C. You, J. Zhang, K. Huang, and K. B. Letaief, "A Survey on Mobile Edge Computing: The
Communication Perspective," IEEE Communications Surveys & Tutorials, vol. 19, no. 4, pp. 2322-2358, 2017.
[3]. F. Bonomi, R. Milito, J. Zhu, and S. Addepalli, "Fog Computing and Its Role in the Internet of Things," in
Proceedings of the First Edition of the MCC Workshop on Mobile Cloud Computing, 2012, pp. 13-16.
[4]. K. Zhang, S. Ren, S. Ruj, B. H. Kim, and K. C. Seo, "SecureThings: A Lightweight and Flexible IoT Service
Framework for Smart Cities," IEEE Transactions on Industrial Informatics, vol. 15, no. 12, pp. 7996-8003, 2019.
[5]. M. Zohrevand, M. Hassanpour, M. Dehghan, and S. F. Shahandashti, "Towards Blockchain-Based Edge
Computing," IEEE Transactions on Industrial Informatics, vol. 16, no. 6, pp. 4017-4025, 2020.

Copyright to IJARSCT DOI: 10.48175/IJARSCT-12921 120

www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)
International Open-Access, Double-Blind, Peer-Reviewed, Refereed, Multidisciplinary Online Journal
Impact Factor: 7.301 Volume 3, Issue 1, September 2023

[6]. Beloglazov, R. Buyya, Y. C. Lee, and A. Zomaya, "A Taxonomy and Survey of Energy-Efficient Data Centers and
Cloud Computing Systems," Advances in Computers, vol. 82, pp. 47-111, 2011.
[7]. Y. Wang, M. Han, C. Zhang, and K. Guan, "Deep Learning for IoT Big Data and Streaming Analytics: A Survey,"
IEEE Access, vol. 7, pp. 1-1, 2019.
[8]. D. Puthal, N. Malik, S. Mohanty, E. Kougianos, and G. Das, "Everything You Wanted to Know About Smart
Cities: The Internet of Things Is the Backbone," IEEE Consumer Electronics Magazine, vol. 7, no. 4, pp. 60-70, 2018.
[9]. A. V. Dastjerdi and R. Buyya, "Fog Computing: Helping the Internet of Things Realize Its Potential," Computer,
vol. 49, no. 8, pp. 112-116, 2016.
[10]. D. Soldani, M. Dohler, N. Shah, and S. D. Silva, "MistIoT: A Decentralized Fog Computing Platform to Support
IoT Applications," IEEE Internet of Things Journal, vol. 6, no. 2, pp. 2074-2087, 2019.

Copyright to IJARSCT DOI: 10.48175/IJARSCT-12921 121

www.ijarsct.co.in