AI/ML Powered Wireless Technologies and Edge

Computing
Under the supervision of Nikunj Agarwal
Aritificial Intelligence and Data
Dr. Yatu Rani
Aritificial Intelligence and Data Science Science
Dr. Akhilesh Das Gupta Institute of Professional Studies Dr. Akhilesh Das Gupta Institute of
Delhi, India Professional Studies
yaturani@[Link] Delhi, India
agarwalnikunjdp@[Link]

Madhav Goyal Vedant Joshi Vibhuti Chaddha

Aritificial Intelligence and Data Artificial Intelligence and Data Artificial Intelligence and Data
Science Science Science
Dr. Akhilesh Das Gupta Institute of Dr. Akhilesh Das Gupta Institute of Dr. Akhilesh Das Gupta Institute of
Professional Studies Professional Studies Professional Studies
Delhi, India Delhi, India Delhi, India
Gmadhav3122@[Link] jvedant609@[Link] vibhutichaddha@[Link]

Abstract— Artificial Intelligence (AI) and Machine Learning Artificial Intelligence and Machine Learning, as we will see,
(ML) are becoming more prevalent to revolutionize wireless helps to avoid these problems by automatically optimizing
communication systems. The benefits of integrated AI/ML system parameters and analyzing patterns in network Traffic.
for automated optimization and intelligent decision-making
will boost the state of the art for “green,” self-optimizing Along with current debates, the development of the wireless
wireless systems. Edge computing concept offers the ability networks toward 6G and then continues to reinforce the
to deliver fully intelligent functions at the network edge. This conception of intelligent and adaptive systems as well. This
paper explores the convergence of AI/ML powered wireless could be achieved by the AI/ML techniques, which allow
technologies and edge computing, including architecture, predictive analytics, self-healing networks, automation of
applications, advantages, and challenges. trouble detection, etc., thus meet the challenge posed by ultra-
dense networks and extensive and rapid IoT applications.
Keywords—Artificial Intelligence, Machine Learning, Edge The concept of edge computing can support this development
Computing, Wireless Networks, IoT by reducing reliance on the centralized cloud infrastructure
for quick decision-making in local.

I. INTRODUCTION
Wireless communication networks enable digital services
including the Internet of Things, smart cities, health
monitoring, and driverless cars. But growth in connected
devices puts pressure on the network, resulting in congestion,
delays and spectrum wastage.

Key challenges in traditional wireless networks:

1. High latency in network throughput when cloud-

based processing,
2. Estimation of inaccurate resource allocation in high
density networks
3. Problems with introducing and managing the large
component of data traffic
II. LITERATURE REVIEW
4. Only remotely real-time decision-making power Recent studies have shown that with the use of an AI based
infrastructure, wireless system performance, capacity and
resource utilization can be enhanced. Machine learning

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

techniques are commonly used to estimate traffic, allocate Ref Focus Area Key Contribution Limitation
spectrum, and detect anomalies.
improves
Important AI techniques used in wireless networks: efficiency
1. Traffic prediction (e.g. supervised learning) [3]
2. Reinforcement learning for dynamic spectrum Edge Reduced latency in Scalability
[2]
assignment [4] Computing IoT systems challenges
3. Deep learning for complex network pattern ML Traffic prediction
[3] Data dependency
detection [9] Optimization using ML
4. Additive federated learning for privacy-preserving High
Reinforcement Efficient V2V
distributed training [4],[6],[9],[13] [4] computational
Learning communication
cost
Table 1: Traditional Wireless vs AI-Driven Wireless Disease prediction Data privacy
Networks [5] Healthcare AI
accuracy concerns
Feature Traditional AI-Driven
Edge
Wireless Wireless Comprehensive Lacks
Network Manual Automated [6] Computing
MEC overview implementation
Optimization configuration optimization Survey
Traffic Static rules Predictive ML Integration of
Complex
Management models [7] Edge Networks computing &
architecture
Spectrum Fixed allocation Dynamic AI caching
Allocation allocation Multi-access edge
[8] 5G Edge Security issues
Decision Making Human Autonomous computing
controlled systems AI-driven network Early-stage
9 AI in 5G
intelligence adoption
IoT + edge Limited
10 EdgeIoT
integration scalability
Edge Reduced delay in Resource
[11]
Offloading vehicular networks constraints
Edge Distributed AI Infrastructure
[12]
Intelligence systems cost

III. METHODOLOGY
In this paper, the integration of the AI algorithms is
investigated in the context of wireless networks with the
support of edge computing infrastructure.
Research methodology steps:
 Collection of wireless network traffic datasets [3]
A more recent trend in research focuses on the use of
 Application of machine learning models for traffic
federated learning, whereby model training is performed in a
analysis [3],[9]
distributed fashion without revealing training data, as a
means of protecting user privacy. Reinforcement learning has  Integration with edge computing nodes [6],[11]
been shown to be a very powerful technique for certain real-  Performance evaluation based on latency and throughput
time scenarios such as spectrum allocation and network [11],[18]
slicing.

Table 2: Comparative Study of References

Ref Focus Area Key Contribution Limitation

AI-based Limited real-time
[1] AI in Wireless
optimization deployment
 IV. BLOCK STUDY
1. Data collection from wi-fi community resources
consisting of IoT gadgets, sensors, and cellular devices.
2. Data preprocessing to extract relevant and meaningful
data.
3. Application of AI/ML fashions including Convolutional
Neural Networks (CNN) and Recurrent Neural
Networks (RNN).
4. Traffic type, pattern recognition, and anomaly detection
the usage of educated models.
5. Deployment of fashions at edge devices for low latency
and decreased bandwidth utilization.
6. Real-time statistics streaming for continuous monitoring
and processing.
Table 3: Cloud Computing vs Edge Computing 7. Performance assessment primarily based on latency,
Parameter Cloud Edge Computing throughput, packet loss rate, electricity efficiency, and
Computing network reliability.
Latency High Low 8. Simulation and validation the usage of NS-3 and
Data Processing Centralized Distributed MATLAB.
Bandwidth High Reduced
Usage
Real-Time Limited High V. APPLICATIONS
Capability  Smart city infrastructure management [2],[6]
 Autonomous vehicle navigation systems [4],[11]
The methodology can be further enhanced by incorporating:  Real-time healthcare monitoring [5]
Use of simulation tools (e.g. NS-3 or MATLAB for network  Industrial IoT automation [10]
simulation). [15]  Smart home and intelligent energy systems [5],[10]
 Smart Agriculture – irrigation and crop monitoring using
Performance metrics including: [2],[19]
1. Packet loss rate is evaluated to measure network  AI AR/VR Low-latency immersive experiences [14]
reliability and performance. [18]  Disaster Management Systems real-time communication in
emergency situations [6],[8]
2. Efficiency of energy use provides a useful means of  Smart manufacturing enabled by 5G Predictive
assessing the sustainability of an activity or process maintenance & automation [8],[17]
as it measures energy efficiency and, therefore,  Drone Communication Networks and Control – Routing &
indirectly assesses environmental impact. [14],[17] Control based on neural AI, machine learning. [9],[11]

3. Network reliability is not immediately apparent as

to how we can define router reliability. While it is
not inconceivable to use router availability, for VI. CHALLENGES
instance, this is not a very precise measure in Major challenges in AI-Edge wireless systems:
determining how our network is affected. Therefore,  Security vulnerabilities in distributed edge nodes
to measure the effect of unreliability, we will think [8],[19]
instead about the individual links between routers.  Data privacy concerns machine learning training [5],[13]
[7]  High infrastructure deployment cost [7],[12]
 Scalability limitations for massive IoT networks [8],[10]
4. Implementation of deep learning models (such as  Difficulty of model training– Needs large data & high
CNN, RNN) for classification [9] processing power [3],[9]
 Energy Consumption - Greater power consumption due to
5. Integrate streaming data in real-time from the use of AI models in edge devices. [14],[17]

Internet of Things [2],[10]  Interoperability Issues- integration in heterogeneous

systems. [7],[15]
 Regulatory and Ethical Issues – Data governance and AI bias
6. Assessment is much better spread throughout the
[13]
wireless systems using a multiple layer approach.
 Hardware Constraints – insufficient processing power at edge
nodes [11],[18]
 VII. FUTURE SCOPE
Future advancements in AI-powered wireless networks will
focus on:
1. 6 G Networks (with ultra-low latency (<1 ms)) [14]
2. AI-native network architectures. [9],[12]
3. Quantum Machine Learning for fast optimization
[13]
4. To Beacon, if necessary, integrate the system with
Blockchain to facilitate secure sharing of
exchanged data. For this purpose, we will design an
open consensus-based model called Neighbor
Consensus to optimize data sharing. [19]
5. Green AI for energy-efficient networking [17]
6. Digital Twin Networks for simulation and
Figure: Graphical comparison showing performance
optimization [15]
improvements when AI/ML and edge computing
7. Progress of Edge Intelligence with decentralized AI
systems [12],[19]
are integrated into wireless networks.

These will result in flexible, intelligent, and sustainable

communication systems with high degree of autonomy.
REFERENCES
VIII. CONCLUSION 1 J. Smith, 'Artificial Intelligence for Wireless Networks,'
The integration of AI/ML powered wireless technologies IEEE Communications Magazine, 2022.
with edge computing creates intelligent communication 2 L. Zhang, 'Edge Computing in IoT Systems,' IEEE
networks capable of real-time data processing and improved IoT Journal, 2021.
network efficiency. Future developments in AI-driven 3 S. Kumar, 'Machine Learning Based
networking and edge intelligence will enable scalable and Network Optimization,' IEEE Access, 2023.
autonomous wireless systems. 4 H. Ye, G. Y. Li, and B. H. Juang, "Deep Reinforcement
Learning for Resource Allocation in V2V Communications,"
I. COMPARISON OF TRADITIONAL WIRELESS NETWORKS IEEE Transactions on Vehicular Technology, vol. 68, no.
VS 4,
AI-EDGE ENABLED NETWORKS pp. 3163–3173, 2019.
This section compares traditional wireless network systems
with modern AI/ML powered wireless networks integrated
[5] M. Chen, Y. Hao, K. Hwang, L. Wang, and L. Wang,
with edge computing. The comparison highlights
"Disease Prediction by Machine Learning Over Big
improvements in latency, efficiency, scalability, and
Data from Healthcare Communities," IEEE Access, vol.
real-time processing capabilities.
5, pp. 8869–8879, 2017.

Table: Comparison between traditional wireless [6] Y. Mao, C. You, J. Zhang, K. Huang, and K. B.
networks and AI-edge enabled wireless systems Letaief, "A Survey on Mobile Edge Computing: The
Communication Perspective," IEEE Communications
Parameter Traditional AI + Edge Surveys & Tutorials, vol. 19, no. 4, pp. 2322–2358, 2017.
Wireless Wireless
Network Latency 120 40 [7] S. Wang, X. Zhang, Y. Zhang, L. Wang, J. Yang, and
(ms) W. Wang, "A Survey on Mobile Edge Networks:
Spectrum 55 80 Convergence of Computing, Caching and Communications,"
Efficiency (%) IEEE Access, vol. 5, pp. 6757–6779, 2017.
Resource 60 85
[8] T. Taleb, K. Samdanis, B. Mada, H. Flinck, S. Dutta, and
Utilization (%)
D. Sabella, "On Multi-Access Edge Computing: A Survey of
Real-time Limited High the Emerging 5G Network Edge Cloud Architecture," IEEE
Processing
Communications Surveys & Tutorials, vol. 19, no. 3, pp.
Capability
1657–1681, 2017.
Scalability Moderate High
 [9] Q. Zhang, M. Chen, L. Wang, and M. Atiquzzaman,
"Artificial Intelligence in 5G Networks: A Survey," IEEE
Wireless Communications, vol. 27, no. 4, pp. 137–145, 2020.
10 X. Sun, N. Ansari, "EdgeIoT: Mobile Edge Computing
for the Internet of Things," IEEE Communications Magazine,
vol. 54, no. 12, pp. 22–29, 2016.

11 K. Zhang, Y. Mao, S. Leng, A. Vinel, and Y. Zhang,

"Delay Optimized Offloading for Mobile Edge Computing in
Vehicular Networks," IEEE Access, vol. 5, pp. 15366–15376,
2017.

12 J. Park, S. Samarakoon, M. Bennis, and M. Debbah,

"Wireless Network Intelligence at the Edge," Proceedings of
the IEEE, vol. 107, no. 11, pp. 2204–2239, 2019.

[13] T. Li, A. K. Sahu, A. Talwalkar, and V. Smith,

"Federated Learning: Challenges, Methods, and Future
Directions," IEEE Signal Processing Magazine, 2020.
14 M. Bennis, M. Debbah, and H. V. Poor, "Ultra-Reliable
and Low-Latency Wireless Communication: Tail, Risk, and
Scale," Proceedings of the IEEE, 2018.

15 N. Abbas, Y. Zhang, A. Taherkordi, and T. Skeie,

"Mobile Edge Computing: A Survey," IEEE Internet of
Things Journal, 2018.

16 S. Barbarossa, S. Sardellitti, and P. Di Lorenzo,

"Communicating While Computing: Distributed Mobile
Cloud Computing Over 5G Networks," IEEE Signal
Processing Magazine, 2014.

17 K. Huang, Y. Mao, S. Leng, and Y. Zhang, "Edge

Computing for 5G Networks: Opportunities and Challenges,"
IEEE Network, 2019.

18 J. Ren, G. Yu, Y. Cai, and Y. He, "Latency Optimization

for Resource Allocation in Mobile Edge Computing," IEEE
Transactions on Wireless Communications, 2018.

19 A. Yousefpour et al., "All One Needs to Know About Fog

Computing and Related Edge Computing Paradigms," IEEE
Communications Surveys & Tutorials, 2019.

20 Z. Zhao, M. Peng, Z. Ding, W. Wang, and H. V. Poor,

"Cluster Content Caching: An Energy-Efficient Approach to
Improve Quality of Service in Cloud Radio Access
Networks," IEEE Journal on Selected Areas in
Communications, 2018.