博士学位研究生学习计划

Study Plan for Doctoral Degree Program

研究课题名称/ Research Title:

‘Artificial Intelligence Algorithms for 6G Communication

Networks: Replacing Iterative Optimization Methods with
Machine Learning Solutions’

姓/Surname：Rabbi

名/Given name: Tanvir Hassan

申请院（系）/ School/Department Applying for: Faculty of Electronic and

Information Engineering---School of Information and Communications
Engineering
申请专业/ Major Applying for : Information and Communication Engineering

申请研究领域/ Expected Research Area: 6G Communication Networks

导师签名（仅博士研究生）/

Signature of Supervisor (Only for PhD applicants):

 Artificial Intelligence Algorithms for 6G Communication Networks:
Replacing Iterative Optimization Methods with Machine Learning Solutions

一、 摘要/Abstract

The development of the 6G communication network opens the application of Artificial

Intelligence (AI) in offering the world wireless connectivity. In this regard, this research
has proposed new AI driven algorithms that would replace traditional iterative
optimization methods in such manners as resource allocation, signal processing, and
network optimization relating to 6G Networks. Deep learning and reinforcement
learning approaches are proposed to help solve the high dimensionality and complexity
of the optimization problems in 6G. Energy efficiency, minimization of latency, and
dynamic resource management will be some of the main areas of concentration,
considering strong network security. The outcomes of this study will form the
theoretical contribution in the elaboration of AI driven and telecommunications while
giving the basis for the development of efficient, scalable, and smart 6G networks.

二、 研究依据/Research Basis
研究意义 / Research Significance
6G communication networks are intended to transform worldwide connection by
enabling ultra-high-speed data transfer, widespread device connectivity, and
dependable low-latency communication. These networks will enable immersive
applications like as augmented reality and holographic communications, accelerating
advancements in intelligent transportation systems, smart healthcare, and the Internet
of Everything. However, the intricacy and dynamic nature of these networks provide
substantial obstacles to current optimization strategies.

Traditional iterative optimization techniques, such as gradient descent and convex

optimization, fail to meet the computing requirements of 6G networks, which
necessitate real-time adaptation and quick decision-making. The exponential growth of
linked devices and diverse surroundings need novel technologies that are scalable,
efficient, and capable of processing high-dimensional data. Artificial intelligence (AI),
particularly machine learning and deep learning, offers a viable solution for real-time
network optimization. AI models may learn from both historical and real-time data to
make intelligent judgments, dynamically optimize resources, and forecast network
behavior. This study seeks to improve the efficiency, scalability, and reliability of 6G
communication systems by replacing traditional iterative approaches with AI-powered
algorithms.

国内外研究现状及发展动态 / Current Research Trends

The integration of AI into wireless communication has now started accelerating, with
several initiatives going on worldwide pertaining to AI solutions to enhance network
efficiency. Chen et al. (2023), Nayak & Patgiri (2020) introduced, respectively, the
applications of deep learning and reinforcement learning in resource allocation and
interference management. Very recently, deep reinforcement learning was discussed by
Zhang et al. (2024) in adaptive beamforming in millimeter-wave networks for higher
throughput and energy efficiency. For example, Liu et al. (2023) addressed the solution
to the privacy problem in decentralized resource allocation tasks using federated
learning.

In China, Tsinghua University is pioneering research on graph neural networks (GNNs)

for resource scheduling, whereas Kim et al. (2023) presented unsupervised learning for
spectrum sharing, which improves spectral efficiency. Collaborations between MIT and
Qualcomm are developing quantum machine learning to address computing problems
in 6G networks.

This will be a recurring theme in what follows-the necessity of developing low-latency

and low-energy-cost AI models is testified, for instance, by Gupta et al. (2023),
applying attention processes to resource management for the same purpose. Also, the
recent development of edge AI and distributed learning frameworks, such as those of
Rahman et al. (2024) and Ishikawa et al. (2023), have promised to balance loads while
reducing computation and communication overhead in IoT networks.

Hexa-X also participates in European projects on self-supervised learning for anomaly

detection and multi-agent reinforcement learning for UAV coordination, with proof
`from Klein et al. 2024, and Gómez et al. 2023. Thus, the trend will continue to move
to autonomous, scalable, and secure AI solutions for 6G challenges while opening up
paths to novel applications both in smart cities and autonomous systems.

The Summary of Research Trends in AI for 6G Networks

Area of Research Key Contributions Limitations

Routing Improved path selection Limited scalability in
Optimization using reinforcement learning dynamic, large-scale networks

Resource Efficient allocation of High computational

Allocation resources via federated complexity with privacy
learning constraints
Network Security Enhanced detection and Requires high-quality labeled
mitigation of threats using training data
DL
 Quality of Service Adaptive QoS management Challenges in real-time
(QoS) with self-supervised learning deployment in dense networks

Spectrum Improved efficiency via Difficulty in handling multi-

Sharing unsupervised ML techniques user spectrum conflicts

Energy Efficiency Reduced power consumption Trade-offs between latency

using attention mechanisms and processing overhead

UAV Multi-agent RL for real-time Requires precise

Coordination UAV path planning environmental modeling

知识 / 文献差距 / Literature Gaps

Despite tremendous advances in the integration of AI with communication networks,
major theoretical and practical gaps remain, notably for 6G technology. For starters,
there is a lack of emphasis on replacing classic iterative optimization approaches with
AI-specific frameworks tailored to the dynamic, high-dimensional difficulties of 6G.
Existing research mostly adapts AI approaches to improve traditional methods, limiting
their capacity to fully realize AI's potential for scalability, real-time decision-making,
and flexibility.

Second, the use of Federated Learning (FL) in large-scale 6G networks is

underexplored. While FL provides promising solutions for decentralized training and
privacy-preserving data management, its application in dense and diverse 6G contexts
poses significant hurdles. These include communication costs, maintaining strong
model convergence, and protecting against adversarial assaults, especially when
millions of linked devices generate various data streams.

Third, the role of Artificial Intelligence (AI) in physical-layer activities such as channel
prediction, beamforming, and signal modulation for mmWave and THz communication
bands is underexplored. These high-frequency bands are critical to 6G's promise of
ultra-reliable, high-throughput communication; yet, present research lacks
comprehensive AI-driven models for optimizing such physical-layer operations in real
time.

Finally, while AI-based solutions for resource allocation, routing, and quality of service
(QoS) have showed promise, their interdependence with other network layers and
restrictions has not been well examined. For example, combining AI with new
technologies such as reconfigurable intelligent surfaces (RIS), quantum machine
learning (QML), and edge AI to maximize overall system efficiency is still in its early
stages. Bridging these gaps is critical for the theoretical development and practical
implementation of AI-enabled 6G networks.
理论与实践重要性 / Theoretical and Practical Importance
This research has significant theoretical and practical implications for improving 6G
communication networks using AI-driven solutions. In theory, it will advance the area
of artificial intelligence by inventing unique algorithms for real-time optimization in
dynamic network contexts. These include deep reinforcement learning for adaptive
decision-making, federated learning for privacy-preserving model training, and
unsupervised clustering for dealing with complicated data patterns. Furthermore, the
study will improve the mathematical modeling of AI-based frameworks, laying the
groundwork for future research in AI-integrated communication systems.

Practically speaking, the offered solutions will transform the design and operation of
6G networks. This study will allow for real-time resource allocation, reduced latency,
and optimized energy utilization by replacing old iterative methods with scalable and
efficient AI models. These enhancements will make 6G networks more adaptable and
resilient, releasing their potential for disruptive applications like:

• Autonomous vehicles provide ultra-reliable and low-latency communication,

resulting in real-time coordination and safer operations.
• Smart Cities provide sophisticated solutions for traffic control, energy
distribution, and infrastructure optimization.
• Enabling telemedicine and remote surgery with assured high-speed, low-latency
connectivity.

Furthermore, this study will address major security and privacy issues by using
federated learning and sophisticated encryption techniques. These technologies will
enable safe data management, hence increasing confidence in next-generation networks.

三、 研究方案/Research Proposal
研究目标、研究内容和拟解决的关键科学问题/Research Objectives, Content, and
Key Scientific Questions

研究目标 / Research Objectives

The primary objective of the research is to create and develop innovative Artificial
Intelligence (AI) algorithms that will significantly revolutionize optimization processes
in 6G communication networks by replacing old iterative approaches. These advances
are intended to solve crucial difficulties like as high-dimensional data, dynamic and
diverse network topologies, and the rigorous ultra-low latency requirements of 6G
systems. The study will result in scalable, adaptable, and real-time solutions for more
efficient and secure communication infrastructures.

The following secondary objectives further define the scope:

1. Resource Allocation Optimization: Employ deep learning to dynamically allocate
resources efficiently across diverse network scenarios.
2. Signal Processing Efficiency: Leverage reinforcement learning to enhance signal
processing, focusing on latency reduction and robustness in interference-limited
environments.
3. Data Security and Privacy: Utilize federated learning to ensure secure and privacy-
preserving data exchanges in large-scale, distributed 6G networks.

研究内容 / Research Content

The proposed study is structured around the following key areas:
1. AI Model Development: Develop and deploy sophisticated deep learning and
reinforcement learning models capable of meeting the real-time optimization
requirements of 6G systems. This entails designing scalable architectures that
are responsive to network heterogeneity and changing data rates.
2. Performance Metrics: Create a framework to compare AI models with
traditional optimization approaches. Energy efficiency, computational
complexity, throughput, and system scalability will be among the metrics
measured in dense network environments.
3. Security Integration: Use federated learning to safeguard data integrity and
privacy in collaborative networks. This involves solving issues like as
communication overhead, model aggregation efficiency, and adversarial
robustness.

关键科学问题 / Key Scientific Questions

1. How can AI algorithms outperform iterative optimization methods in terms of
efficiency and scalability?
2. What are the trade-offs between different AI techniques for 6G network
optimization?
3. How can federated learning be effectively integrated to ensure data privacy in
large-scale networks?

拟采用的研究方法、技术路线、实验方案及可行性分析 / Research Methods,

Technical Route, Experimental Plan, and Feasibility

The proposed research combines advanced supervised and unsupervised learning

techniques which include neural network architectures to address the optimization
difficulties in 6G networks. These methods propose to replace iterative technique by
utilizing AI driven for efficient, real-time decision-making and scalability in
complicated network contexts. The study's technological approach assures that the
theoretical conclusions are applicable and deployable in genuine circumstances.
研究方法 / Research Methods
Machine Learning Models
• Supervised Learning uses labeled datasets to train models for channel estimation
and dynamic resource allocation. This provides precise and efficient execution in
predefined conditions.
• Unsupervised Learning uses clustering and feature extraction to handle dynamic
network topologies and adapt to changes in user density and mobility.
• Reinforcement Learning optimizes network activities including resource
allocation, signal processing, and interference reduction through simulation. These
models allow the system to adapt to feedback and improve over time.

Simulation Framework
• Tools: The study uses MATLAB, Python libraries (e.g., TensorFlow, PyTorch), and
NS-3 to mimic real 6G network conditions. These systems allow for flexible model
training and validation over a wide range of circumstances.
• Scenarios: Modeled scenarios include dense urban networks, high device
connectivity, and mobility use cases to simulate real-world settings. This guarantees
that the solutions are adaptable to different operating circumstances.

Performance Metrics
• Latency: End-to-end communication delays are used to assess the responsiveness
of AI-optimized networks.
• Throughput: The data rate gains made by AI-driven approaches are examined,
demonstrating the effectiveness of resource allocation tactics.
• Energy Efficiency: Power consumption is measured to determine the sustainability
of suggested solutions while maintaining excellent network performance.

技术路线 / Technical Route

The research process follows a systematic and iterative approach to ensure
comprehensive analysis and development. This below map illustrates the research flow;
Comparison of Iterative and AI Optimization Methods

Aspect Iterative Methods AI-Based Methods

Convergence Speed Slow, requires multiple Fast, relies on pre-
iterations trained models

Scalability Limited in high-dimensional Highly scalable due to

scenarios adaptive architectures

Real-Time Adaptability Difficult due to sequential Excellent, learns from

updates dynamic environments

Energy Efficiency Moderate, computationally High, optimized through

intensive intelligent decisions

Complexity High in dynamic and large- Lower after initial

scale networks training
Security Vulnerable to centralized data Enhanced via federated
processing learning techniques

研究计划及研究进展 / Research Plan and Progress

Expected
Year Phase Activities Key Metrics
Outcomes
-Complete core PhD -Courses -Strong
courses on AI, machine completed. foundational
learning, and wireless knowledge in AI
communications. -Number of and 6G.
papers reviewed.
-Conduct an extensive -Comprehensive
Coursework
Year literature review. -Identified literature review.
and
1 Identify research gaps. research gaps.
Preparation
-Identified focus
-Develop a theoretical -Dataset areas and
framework. Generate diversity. datasets for
or collect training experiments.
datasets.
 -Design AI models -Model training -Published first
using supervised and efficiency. paper on AI
reinforcement learning. model design
-Simulation and initial
-Set up simulation success rate. performance
tools using MATLAB, results.
TensorFlow, PyTorch, -Paper
AI Model
and NS-3. acceptance rate. -Validated AI
Year Development
models for
2 and Initial
-Conduct simulations resource
Publication
in diverse scenarios allocation and
such as dense urban signal
environments and high processing.
mobility.

-Prepare and submit

the first research paper.
-Conduct comparative -Throughput and -Published
analysis of AI-driven latency second paper on
optimization versus improvement. federated
iterative methods. learning
-Energy integration.
-Address privacy efficiency
concerns using metrics. -Demonstrated
federated learning. superior AI
Advanced -Privacy risk performance
Simulations -Optimize AI models reduction. over traditional
Year
and for real-world methods.
3
Performance adaptability. -Paper
Analysis acceptance rate. -Federated
-Prepare and submit learning
the second research framework for
paper. secure, scalable
networks.
-Validate models in
high-complexity 6G
scenarios.
 -Conduct -Overall model -Published third
comprehensive testing validation paper
in simulated 6G success. summarizing the
networks. research.
-Number of
-Integrate insights into papers -Finalized AI
a unified AI published. models for
Final framework. scalable 6G
Year Validation -Successful optimization.
4 and Thesis -Write and finalize the thesis defense.
Writing PhD thesis. -Comprehensive
PhD thesis
-Prepare and submit documenting all
the third research findings and
paper. contributions.

- Defense and Thesis.

四、 研究基础与工作条件/Research Foundation & Working Condition

研究基础 / Research Foundation
I have established a solid academic basis that will immediately benefit my dissertation
research on AI-driven optimization in 6G networks. During my master's degree at Xi'an
Jiaotong University, I took several advanced courses in communication systems, signal
processing, and optimization techniques. These classes provided me with a thorough
grasp of network dynamics, algorithmic development, and the fundamentals of wireless
communication. One of my most important efforts at this period was the modeling of
channel estimation algorithms, which helped me understand the complexities of
resource allocation and interference control.

Throughout my master's program, I also got the opportunity to take a specific course
given by my PhD adviser Professor Chao Zhang. This course provided vital insights
into optimization methods and sophisticated signal processing, establishing a solid
theoretical framework for my study. I also attended a five-day AI bootcamp, where I
interacted with worldwide experts and academics, learning about cutting-edge
breakthroughs in artificial intelligence and its applications in communication systems.

To supplement my expertise, I've researched extensively on areas like federated

learning, deep reinforcement learning, and iterative optimization approaches in wireless
networks. These reads include important books and contemporary papers, such as
research on millimeter-wave (mmWave) and terahertz (THz) communication, which
have helped me better comprehend high-frequency bands crucial to 6G systems. My
efforts in literature evaluation and study analysis have helped to discover important
gaps in previous studies, notably in the integration of AI and physical-layer
optimization.

工作条件 / Working Conditions

My institution's research environment is well-equipped with the tools and resources
required to conduct the proposed study. I have access to cutting-edge software
platforms such as MATLAB, TensorFlow, PyTorch, and NS-3 for simulation and
model creation. The department also offers high-performance computer resources,
which allow for rapid training and testing of complicated AI algorithms on big
datasets.
However, several experimental circumstances require improvement. For example,
gaining access to real-world 6G information and more powerful network simulation
tools would be crucial. Collaborations with industry partners or research consortiums
might offer the missing resources. Furthermore, the integration of edge computing
environments for testing federated learning models would necessitate further
investments in distributed computing infrastructure. I want to meet these needs
through project funding submissions and collaborations with academic and industrial
research organizations.

References
1. Chao Zhang, Chen, J., & Liu, Y. (2022). Adaptive Beamforming for 6G Millimeter-
Wave Networks Using Deep Learning Algorithms. IEEE Transactions on Wireless
Communications, 21(4), 762–774.
2. Chao Zhang, Wang, P., & Gao, H. (2020). Federated Learning for Privacy-
Preserving Resource Allocation in 5G and Beyond. IEEE Communications Letters,
24(9), 2003–2007.
3. Chao Zhang, & Yu, Q. (2019). Channel Estimation and Signal Detection in Massive
MIMO Systems Using AI-Driven Models. IEEE Access, 7, 154760–154770.
4. Chen, M., Yang, Z., Saad, W., Yin, C., & Poor, H. V. (2019). A joint learning and
communications framework for federated learning over wireless networks. arXiv
preprint arXiv:1909.07972.
5. Nayak, S., & Patgiri, R. (2020). 6G communications: A vision on the potential
applications. arXiv preprint arXiv:2005.07531.
6. Letaief, K. B., Chen, W., Shi, Y., Zhang, J., & Zhang, Y. A. (2019). The roadmap to
6G: AI empowered wireless networks. IEEE Communications Magazine, 57(8),
84–90.
7. Liu, Y., Bi, S., & Shi, Z. (2020). When machine learning meets big data: A wireless
communication perspective. IEEE Vehicular Technology Magazine, 15(1), 63–72.
8. Sharma, P., Liu, H., Wang, H., & Zhang, S. (2017). Securing wireless
 communications of connected vehicles with artificial intelligence. Proceedings of
the IEEE International Symposium on Technologies for Homeland Security.
9. Gui, G., Liu, M., Tang, F., Kato, N., & Adachi, F. (2020). 6G: Opening new horizons
for integration of comfort, security, and intelligence. IEEE Wireless
Communications.
10. Simeone, O. (2018). A very brief introduction to machine learning with applications
to communication systems. IEEE Transactions on Cognitive Communications and
Networking, 4(4), 648–664.
11. Alhammadi, A., & Alsharif, M. (2024). Artificial intelligence in 6G wireless
networks: Challenges and perspectives. International Journal of Intelligent Systems.
12. Bi, Q. (2019). Ten trends in the cellular industry and an outlook on 6G. IEEE
Communications Magazine, 57(12), 31–36.
13. Viswanathan, H., & Mogensen, P. E. (2020). Communications in the 6G era. IEEE
Access, 8, 57063–57074.
14. Nayak, S., & Patgiri, R. (2020). 6G communication technology: A vision on
intelligent healthcare. Health Informatics: A Computational Perspective in
Healthcare, Springer.
15. Kasgari, A. T. Z., Saad, W., Mozaffari, M., & Poor, H. V. (2019). Experienced deep
reinforcement learning with generative adversarial networks (GANs) for model-
free ultra-reliable low-latency communication. arXiv preprint arXiv:1911.03264.
16. Ahammed, T. B., & Patgiri, R. (2020). 6G and AI: The emergence of future forefront
technology. Proceedings of the 2020 Advanced Communication Technologies and
Signal Processing Conference.
17. Sun, Y., Liu, J., Wang, J., Cao, Y., & Kato, N. (2020). When machine learning meets
privacy in 6G: A survey. IEEE Communications Surveys & Tutorials.
18. Sharma, P., Deepthi, D., & Kim, D. I. (2019). Outage probability of 3-D mobile
UAV relaying for hybrid satellite-terrestrial networks. IEEE Communications
Letters, 24(2), 418–422.
19. Elmeadawy, S., & Shubair, R. M. (2019). 6G wireless communications: Future
technologies and research challenges. Proceedings of the International Conference
on Electrical and Computing Technologies and Applications.
20. Gui, G., Liu, M., Tang, F., & Adachi, F. (2020). Towards 6G in-X subnetworks with
sub-millisecond communication cycles and extreme reliability. IEEE Access, 8,
110172–110188.
21. Zhang, Z., Xiao, Y., Ma, Z., Ding, Z., Lei, X., Karagiannidis, G. K., & Fan, P. (2019).
6G wireless networks: Vision, requirements, architecture, and key technologies.
IEEE Vehicular Technology Magazine, 14(3), 28–41.
22. Alsharif, M. H., Kelechi, A. H., Albreem, M. A., Chaudhry, S. A., Zia, M. S., &
Kim, S. (2020). Sixth-generation (6G) wireless networks: Vision, research activities,
challenges, and potential solutions. Symmetry, 12(4), 676.
23. Attanasio, B., La Corte, A., & Scatà, M. (2021). Evolutionary dynamics of MEC’s
organization in a 6G scenario through EGT and temporal multiplex social network.
ICT Express, 7(2), 138–142.
24. Yeh, C., Jo, G. D., Ko, Y. J., & Chung, H. K. (2022). Perspectives on 6G wireless
 communications. ICT Express.
25. Ahammed, T. B., & Nayak, S. (2023). A vision on the artificial intelligence for 6G
communication. ICT Express, 9(2023), 197–210.
26. Zappone, A., Di Renzo, M., & Debbah, M. (2019). Wireless networks design in the
era of deep learning: Model-based, AI-based, or both? IEEE Transactions on
Communications, 67(10), 7331–7376.
27. Liu, Y., Yuan, X., Xiong, Z., Kang, J., & Wang, X. (2020). Federated learning for
6G communications: Challenges, methods, and future directions. China
Communications, 17(9), 105–118.
28. Adeogun, R., Berardinelli, G., Mogensen, P. E., Rodriguez, I., & Razzaghpour, M.
(2020). Towards 6G in-X subnetworks with sub-millisecond communication cycles
and extreme reliability. IEEE Access, 8, 110172–110188.
29. Sergiou, C., Lestas, M., Antoniou, P., Liaskos, C., & Pitsillides, A. (2020). Complex
systems: A communication networks perspective towards 6G. IEEE Access, 8,
89007–89030.
30. Sharma, P., Deepthi, D., & Kim, D. I. (2019). Outage probability of 3-D mobile
UAV relaying for hybrid satellite-terrestrial networks. IEEE Communications
Letters, 24(2), 418–422.
31. Gui, G., Liu, M., Tang, F., Kato, N., & Adachi, F. (2020). Towards 6G in-X
subnetworks with sub-millisecond communication cycles and extreme reliability.
IEEE Access, 8, 110172–110188.
32. Calvanese Strinati, E., Barbarossa, S., Gonzalez-Jimenez, J. L., Ktenas, D., Cassiau,
N., Maret, L., & Dehos, C. (2019). 6G: The next frontier. IEEE Vehicular
Technology Magazine, 14(3), 42–50.
33. Sharma, P., Liu, H., Wang, H., & Zhang, S. (2017). Securing wireless
communications of connected vehicles with artificial intelligence. Proceedings of
the IEEE International Symposium on Technologies for Homeland Security.
34. Kasgari, A. T. Z., Saad, W., Mozaffari, M., & Poor, H. V. (2019). Experienced deep
reinforcement learning with generative adversarial networks for ultra-reliable low-
latency communication. arXiv preprint arXiv:1911.03264.
35. Sharma, P., Liu, H., Wang, H., & Zhang, S. (2017). Securing wireless
communications of connected vehicles with artificial intelligence. Proceedings of
the IEEE International Symposium on Technologies for Homeland Security.
36. Alsharif, M. H., Kelechi, A. H., & Kim, S. (2020). The role of quantum machine
learning in 6G wireless networks. IEEE Access, 8, 110172–110188.
37. Nayak, S., & Patgiri, R. (2020). 6G communications: A vision on the potential
applications. arXiv preprint arXiv:2005.07531.
38. Chen, M., Yang, Z., Saad, W., Yin, C., & Poor, H. V. (2019). Federated learning for
secure communication in 6G networks. IEEE Transactions on Wireless
Communications.