DEVELOPMENT OF A CNN-BASED FRAMEWORK FOR FAULT DETECTION AND
LOCALIZATION IN FIBER OPTIC CABLES
BY
ADEKOYA AYOMIDE DAVID
(20CK027585)
JUNE 2025
DEVELOPMENT OF A CNN-BASED FRAMEWORK FOR FAULT DETECTION AND
LOCALIZATION IN FIBER OPTIC CABLES
ADEKOYA AYOMIDE DAVID
(20CK027585)
A PROJECT REPORT TO BE SUBMITTED TO THE DEPARTMENT OF
ELECTRICAL & INFORMATION ENGINEERING, IN PARTIAL FULFILMENT OF
THE REQUIREMENTS FOR THE AWARD OF THE BACHELOR OF ENGINEERING
DEGREE (B.Eng) IN INFORMATON AND COMMUNICATION ENGINEERING,
COVENANT UNIVERSITY, OTA, OGUN STATE, NIGERIA.
PROJECT SUPERVISOR
DR. ADEWALE ADEYINKA
JUNE 2025
DECLARATION
I hereby declare that I carried out the work reported in this project in the Department of
Electrical and Information Engineering, Covenant University, under the supervision of Dr.
Adewale Adeyinka. I also solemnly declare that to the best of my knowledge, no part of this
report has been submitted here or elsewhere in a previous application for the award of a degree.
All sources of knowledge used herein have been duly acknowledged.
…………………………………….
(ADEKOYA AYOMIDE DAVID)
20CK027585
CERTIFICATION
This is to certify that the project titled “Development of a CNN-Based Framework for Fault
Detection and Localization in Fiber Optic Cables” by Adekoya Ayomide David, meets the
requirements and regulations governing the award of the Bachelor of Engineering (Information
and Communication Engineering) degree of Covenant University and is approved for its
contribution to knowledge and literary presentation.
Supervisor: Sign: ………………………
Name: Dr. Adewale Adeyinka Date:
Internal Examiner: Sign: ………………………
Name: ……………………. Date:
Head Of Department: Sign: ……………………….
Name: Dr. Isaac Samuel Date:
DEDICATION
ACKNOWLEDGEMENTS
TABLE OF CONTENTS
LIST OF FIGURES
LIST OF TABLES
LIST OF ABBREVIATIONS
OTDR: Optical Time-Domain Reflectometer
CNN: Convolutional Neural Network
ABSTRACT
CHAPTER ONE
INTRODUCTION
1.1. BACKGROUND OF THE STUDY
Reliability of fiber optic networks is essential in modern communications due to high
performance and latency. Defects in fiber optic cables such as breaks, kinks and splices can
cause serious disruptions to data transmission. In general, OTDR technology is used to measure
and detect faults by analyzing the visible signals to find abnormalities in the cable. The OTDR
device works by sending a laser pulse down a fiber optic cable and measuring the return time and
amplitude of the reflected signal. This information is then used to create a signal file that shows
the location and severity of all faults in the cable, including breaks, splices, and attenuation.
OTDR trace files are often interpreted by the Fiber Technicians to identify and correct faults.
However, this traditional method is limited in response time, size, and accuracy, especially for
small fault detection. Recently, deep learning models especially CNNs has emerged as a
powerful tool for signal analysis, due to its ability to automatically extract features from data.
The goal of this project is to use CNNs to develop a more efficient and effective method for
detecting and identifying faults in fiber optic cables, using OTDR trace data as the input.
1.2. SIGNIFICANCE AND MOTIVATION OF THE STUDY
The deployment of an optimized CNN-based framework for fault detection and localization in
fiber optic cables offers numerous advantages. Firstly, such a system improves accuracy by
leveraging CNNs' ability to automatically learn and recognize fault patterns from large datasets,
enhancing the precision detecting and locating faults. This capability minimizes errors that might
arise from manual interpretation of OTDR traces.
Moreover, the system enables real-time analysis by processing OTDR data quickly, allowing
network technicians to identify and respond to faults more rapidly. This prompt detection
significantly reduces the downtime associated with fiber optic network disruptions. Enhanced
accuracy and response times also contribute to improved network reliability, as faults are
detected and pinpointed with higher precision.
Consequently, the operational costs and user inconvenience caused by prolonged outages are
minimized. In addition, the integration of deep learning introduces a high level of automation,
reducing reliance on manual interpretation of OTDR traces. This automation not only simplifies
the diagnostic process but also makes it scalable to larger and more complex fiber optic
networks, enabling efficient fault management across extensive infrastructure.
1.3. PROBLEM STATEMENT
Despite the efficiency of OTDR in identifying faults, the interpretation of OTDR traces requires
expert knowledge, which is both time-consuming and subject to human error. Existing solutions
lack a seamless integration of advanced machine learning models with OTDR systems for
accurate fault detection and localization. This research aims to address the following problem:
How can a CNN-based model be developed and implemented to improve the accuracy and
efficiency of fault detection and localization in fiber optic networks?
1.4. AIM AND OBJECTIVES
Aim:
To develop a robust CNN-based framework for accurate and real-time fault detection and
localization in fiber optic cables using OTDR trace files.
Objectives:
1. To collect and preprocess diverse OTDR trace data representing various fault types and
normal conditions.
2. To design and train a CNN model that can accurately detect and classify fault types
within OTDR traces.
3. To test and validate the model against real world and simulated OTDR data for accuracy,
robustness, and generalization.
4. Create a user-friendly interface that replicates an OTDR device, enabling users to upload
trace files, review predictions, and visualize the results of fault detection and localization.
1.5. METHODOLOGY
1. Literature Review and Planning: Extensive literature review of the existing fault
detection systems to identify the gaps and hence inform the design of the CNN-based
model.
2. Collecting and preprocessing OTDR trace data to train the CNN model:
Collect the OTDR trace data representing various fault types, such as fiber breaks,
bends, and splices. Label data with fault types and locations to support both detection
and localization.
3. Designing a CNN architecture for fault classification and localization: Develop a
custom CNN architecture for fault classification and localization, suitable for processing
OTDR trace images.
4. Model training and validation: The model is trained and validated using a labeled
dataset so that it will make accurate predictions on both fault type and location.
5. Integration into a Streamlit Application: Integrate the trained model into an
interactive application using Streamlit, allowing for user-friendly fault analysis.
6. Reporting: Document all stages of the project, from research and design to
implementation and results. Emphasize challenges and make recommendations for
future improvements.
1.6. SCOPE AND LIMITATION OF THE STUDY
Scope:
The project will focus on developing and testing a CNN-based simulation interface for fault
detection and localization within fiber optic cables, utilizing OTDR trace files as input. The
simulation will be designed to mimic how the model would function on an actual OTDR device,
including the interface for file upload, prediction display, and OTDR trace visualization.
Limitations:
1. Most of the testing of the model's accuracy and effectiveness will be done using pre-
existing OTDR datasets. Testing therefore cannot be fully representative in capturing all
the diversities and complexities of field fiber networks, and this might affect the
generalization of the model performance at field conditions.
2. While the project will be developed on a local machine, assessing real-time processing
and scalability is out of scope. The main attention will be driven towards showing the
functionality of the fault detection model and its accuracy, not towards optimization to
high-speed and extensive data processing.
3. Deep learning models, especially of a hybrid nature of CNN-RNN architectures with
attention mechanisms, do require considerable computational resources in terms of
training.
1.7. PROJECT REPORT ORGANISATION
The Project Report was organized into five chapters:
Chapter One: The Introduction, which outlines the background, significance, problem
statement, aims, objectives, methodology, scope, and limitations of the study.
Chapter Two: The Literature Review looks at related research in the area of fault detection in
fiber optic networks and CNN applications.
Chapter Three: The System analysis and design gives details about the design of the model,
data pre-processing, and the development process of the application.
Chapter Four: The Results and Discussion, which presents the experimental results and
analyzing the model's performance.
Chapter Five: The Conclusion and Recommendations, which summarizes the findings and
proposing future work.
CHAPTER 2:
LITERATURE REVIEW
2.1. INTRODUCTION
