| Module Code | CSP55040 |
| Module Name | Machine Learning Application to Radio and Optical Networks |
| ECTS Weighting [1] | 5 ECTS |
| Semester Taught | Semester 2 |
| Module Coordinator/s | Prof Marco Ruffini and Prof Merim Dzaferagic |
Module Learning Outcomes
On successful completion of this module a student will be able to:
- Identify and analyse key challenges in radio and optical networking, and explore how machine learning algorithms can effectively address these challenges.
- Evaluate the key differences between proprietary and open networks, including their advantages and disadvantages, and assess their implications for network design and management.
- Examine the complexities of data collection in the context of networking, and develop the skills to assess data quality, curate datasets, and ensure they are suitable for machine learning applications
- Design, develop, and implement machine learning algorithms, train them using real-world datasets, and critically assess their performance within radio and optical network environments
- Apply machine learning techniques to optimise network performance, including tasks such as signal to noise ratio prediction, device modeling, quality of transmission estimation, anomaly detection, in radio and optical networks
Module Content
This module focuses on practical application of machine learning techniques to radio and optical transmission networks. It will start with an overview of the machine learning techniques that are applicable to some specific problems in the networking domain and then provide deeper insight into those that will be used in the lab to address the specific use cases described below.
The main focus of the module will be on lab tutorials and exercises, using the python programming language and working with real dataset collected from research testbed infrastructure.
Specific topics addressed in this module include:
- Overview of machine learning techniques applicable to radio and optical networking problems
- Introduction to wireless transmission and networking challenges
- Open-RAN, RAN Intelligent Controller, data collection and preparation
- Application of machine learning to identify different mobility classes from a dataset collected in the RAN.
- Application of machine learning to predict throughput based on RAN performance measurement
- Introduction to optical transmission systems, impairments, challenges and key performance monitor parameters
- Open optical networks, telemetry, data collection and preparation
- Application of machine learning and transfer learning to optical device characterisation
- Application of machine learning to quality of transmission estimation in dynamic optical systems
Teaching and Learning Methods
Teaching and learning will be based on lectures and laboratory tutorials.
Assessment Details
Students must submit a meaningful attempt at a minimum of 80% of the assignments set for this module.
| Assessment Component | Brief Description | Learning Outcomes Addressed | % of Total | Week Set | Week Due |
| O-RAN ML application | Train ML models to classify mobility patterns from RAN datasets and predict network throughput based on RAN performance metrics. | LO3, LO4, LO5 | 40% | 10 | 11 |
| Optical ML application | Train ML models and apply transfer learning for characterizing optical devices and estimating transmission quality in dynamic optical systems. | LO3, LO4, LO5 | 40% | 11 | 12 |
| In class quiz | In class test on application of ML to radio and optical networks | LO1, LO2, LO3 | 20% | n/a | 12 |
Reassessment Details
Assignment (100%)
Contact Hours and Indicative Student Workload
| Contact Hours (scheduled hours per student over full module), broken down by: | 33 hours |
| Lecture | 11 hours |
| Tutorial or seminar | 22 hours |
| Independent study (outside scheduled contact hours), broken down by: | 42 hours |
| Preparation for classes and review of material (including preparation for examination, if applicable) | 22 hours |
| Completion of assessments (including examination, if applicable) | 20 hours |
| Total Hours | 75 hours |
Required Reading List
- R. Ramaswami, K. N. Sivarajan, G. H. Sasaki, Morgan Kaufmann. Optical Networks: A Practical Perspective, 3rd Edition, 2010.
- Eldar, Yonina C., et al., “Machine learning and wireless communications. “, Cambridge University Press, 2022.
- F. N. Khan, et al. Optical Fiber Telecommunications VII, Chapter 21 “Machine learning methods for optical communication systems and networks”, 2020.
Module Pre-requisites
Prerequisite modules: CS7CS4/ CSU44061 Machine Learning and CSU44031Next Generation Networks
Other/alternative non-module prerequisites: General knowledge of networking protocols and transmission.
Module Co-requisites
N/A