All Available Projects

This is a mentored project by IBM, where you can join weekly group calls to brainstorm, ask questions, get ideas, get suggestions, and (to some extent) get technical support / guidance for the IBM technologies used!

Objective: 

Create a chatbot that answers user questions about vehicle health using OBD-II data and IBM Granite. 

Dataset: 

KIT OBD-II Dataset 

IBM Granite Use: 

Power the chatbot’s natural language responses to questions like “What is the status of my engine?” 

MoSCoW Requirements: 

Must: 

• Parse OBD-II logs and extract key metrics 
• Use Granite to answer diagnostics questions 

Should: 

• Support follow-up questions 
• Provide explanations for fault codes 

Could: 

• Integrate with a web or mobile UI 

Won’t: 

• Connect to live vehicle systems 

Datasets for ODB-II data 

1. KIT Automotive OBD-II Dataset 

Source: Karlsruhe Institute of Technology (KIT) 

Link: KIT Dataset on RADAR 

Details: 

CSV format 

Includes: engine RPM, coolant temperature, throttle position, intake air temperature, vehicle speed, MAF, and more 

Logged using KIWI 3 dongle + OBD Auto Doctor app 

Covers various driving conditions (e.g., traffic, highway) 

📁 2. Vdataset Automotive OBD-II Dataset 

Source: Karlsruhe Institute of Technology (duplicate of KIT dataset, alternate host) 

Link: Vdataset Entry 

License: CC BY 4.0 

Details: Same dataset as above, but hosted on a different platform for broader access. 

🧠 3. Kaggle – Search for OBD-II or Vehicle Telemetry 

Source: Kaggle Datasets 

Search Terms: “OBD-II”, “vehicle telemetry”, “car diagnostics” 

Examples: 

“Car Sensor Data” (includes RPM, speed, throttle) 

“Vehicle Sensor Data for Predictive Maintenance” 

“Driving Behavior Dataset” 

Required Learning - https://ibm.biz/SBAICourses  - Open, Start and Review the IBM Granite related badges 

A few important notes:

• the point of contact about this project is available to answer questions and confirm whether it is available for you to take on still
• you still need to approach potential supervisors, who feel able to guide you through this project
• you also need to make this project your own, and determine some details as it is defined at a higher level, which would all go into your project proposal

This project is open for multiple students, though you need to work independently (cannot collaborate on literature review, methodology, development, etc).

This project is open for multiple students, though you need to work independently (cannot collaborate on literature review, methodology, development, etc).

IBM have made available their 7 bit quantum computer online for anyone to access. This project would involve you interacting with the IBM Quantum Computer and programming it to solve a simple problem.

A range of problems can be undertaken such as (these are only indicative)

• Factor a number
• Build a quantum random number generator (maybe)
• Open to other suggestions

Please note that you will need a little bit of mathematical skill to work with this project. I suggest that you log onto the quantum computer and investigate on your own first to see what you can do and whether it is right for you. Quantum Lab - IBM Quantum

A range of simple problems can be undertaken including in discussion with the supervisor such as (these are only indicative)

• Factor a number
• Build a quantum random number generator (maybe)
• Open to other suggestions

Please note that you will need a little bit of mathematical skill to work with this project. I suggest that you log onto the quantum computer and investigate on your own first to see what you can do and whether it is right for you. Quantum Lab - IBM Quantum

Information-centric networking is a new networking paradigm that focuses on content rather than hosts. Ubiquitous in-network caching is one of the key aspects of ICN which has received widespread research interest in recent years. ICN in-network caching has promised to improve various aspects of content delivery in the Internet including reducing bandwidth usage, decreasing content delivery latency, etc.

While in-network caching in ICN has potential to improve content delivery process in the Internet, there have not been a consensus on how it should be done. Various caching strategies / schemes have been proposed in the literature. In this project, the aim is to design, implement and evaluate novel in-network caching algorithms.

Interested student may refer to the following paper to get the context of this project:

https://www.sciencedirect.com/science/article/pii/S014036641300025X

• Student will be given pointers to relevant literature at the beginning of the project.
• Student can freely choose the tool/programming language for this project (python, C/C++, MATLAB, Octave, existing network simulators, etc).
• Student will be given a basic simulation tool (implemented in MATLAB) to kickstart the project.
• This project is open to multiple students. 

This project is associated under the Flexible and Complex Network (FLEXNET) Research Group. 

The ability of a network to withstand perturbations such as link or node failures is important to ensure seamless provision of services. Information-centric networking (ICN) is a new networking paradigm that focuses on content rather than hosts that has potential to improve network resilience. In this project, the focus will be on information resilience in information-centric networks. 

• Student will be given pointers to relevant literature at the beginning of the project.
• Student can freely choose the tool/programming language for this project (python, C/C++, MATLAB, Octave, existing network simulators, etc).
• This project is open to multiple students.

This project is associated under the Flexible and Complex Network (FLEXNET) Research Group. 

The Internet has grown significantly in size and complexity. In this project, the aim is to analyse the evolution of the Internet topology at the domain level including in the considerations their connections, inter-domain relationship, their tiers. 

• Student will be given pointers to relevant literature at the beginning of the project.
• Student will be given the necessary dataset.
• Student can freely choose the tool/programming language for this project (python, C/C++, MATLAB, Octave, existing network simulators, etc).

This project is associated under the Flexible and Complex Network (FLEXNET) Research Group. 

Networks facilitate spreading of various entities (e.g., news, information, viral marketing, virus, malware, etc.). In this project, the student is to build a simulator to simulate and understand spreading processes in networks with different properties/characteristics. 

Interested student may refer to the following paper to get the context of this project (do not worry about the Maths in the paper):

https://ieeexplore.ieee.org/document/7543480

https://ieeexplore.ieee.org/document/8074785

• Student will be given pointers to relevant literature at the beginning of the project.
• Student can freely choose the tool/programming language for this project (python, C/C++, MATLAB, Octave, existing network simulators, etc).

This project is associated under the Flexible and Complex Network (FLEXNET) Research Group. 

The General Data Protection Regulation (GDPR) applies to most businesses in the UK requiring them to safeguard personal data [1] . In the UK, the Data Protection Act 2018  (DPA) was enacted to implement this regulation into law in the UK. This project will look at what steps an organisation must take to respond to a data breach, or suspected data breach to not only ensure compliance with the DPA but also keep the organisation and their customer's data safe. 

This project could, for example, leverage existing frameworks, such as security by design, privacy by design, incident response planning, disaster management or Business Continuity frameworks and/or look to appropriate standards (ISO and/or NIST) for inspiration to create a Data Privacy Incident Response Framework to guide practitioners in handing any data breach or similarly incident in a timely and appropriate manner. 

The General Data Protection Regulation (GDPR) applies to most businesses in the UK requiring them to safeguard personal data[1] . In the UK, the Data Protection Act 2018  (DPA) was enacted to implement this regulation into law in the UK [2].  This project will seek to create best practice guidance for SMEs to follow about what they must to, not only comply, but also demonstrate compliance with DPA and GDPR.

Past literature either focuses on the process of virus spreading in a static network or topology changes. The co-evolution of the both recently received increasing attention especially due to COVID-19. In this project, the aim is to study this coevolution and gain insights into the behaviour of the spreading patterns under different network adaptation strategies. 

• Student will be given pointers to relevant literature at the beginning of the project.
• Student with strong Mathematical ability would have an advantage for this project. 
• Student can freely choose the Mathematical tools or programming language for this project (python, C/C++, Java, MATLAB, Octave, etc).

This project is associated under the Flexible and Complex Network (FLEXNET) Research Group. 

Securing the privacy of organizational and consumer data, and complying with different cyber security requirements, is an ongoing challenge for those providing services at both a local and global scale. Compliance policies specify allowed system behaviour and are ubiquitous in today's IT systems, ranging from access-control and security policies to regulations of personally identifiable data processing.

In this project area, you will have the opportunity to work on developing techniques and tools to support online monitoring of system behaviours with respect to compliance policies. The goal is to providing data stakeholders with ongoing intelligence about the state of their data.

Cyber threat intelligence involves the real-time collection, normalisation, and analysis of the data generated by users, applications and infrastructure and information ingested from external sources - that impacts the IT security and risk posture of an enterprise.

In this project area, you will have the opportunity to work on developing and implementing cyber threat intelligence techniques and tools to provide actionable and comprehensive insight that reduces risk and operational effort for any size organization.

Possible projects include:

1. develop and implement an actionable intelligence platform
2. develop a model for threat landscape understanding
3. develop a framework for threat actor identification, monitoring and intervention
4. develop and implement techniques for threat actor profiling
5. investigation of cyber-criminal ecosystem, their organizational structures and business models

In this project area, you will have the opportunity of investigating the cyber security risk of an organization and providing appropriate cyber security management strategies.

Possible projects include:

1. develop and/or enhance techniques for continuously gauging an organization’s cyber security capability/maturity and recommending improvements
2. investigate and develop a framework for secured application development and deployment
3. develop bespoke cyber security awareness and training procedures that specifically address the work culture and needs of a chosen organization or enterprise

• the point of contact about this project is available to answer questions and confirm whether it is available for you to take on still
• you still need to approach potential supervisors, who feel able to guide you through this project
• you also need to make this project your own, and determine some details as it is defined at a higher level, which would all go into your project proposal

• the point of contact about this project is available to answer questions and confirm whether it is available for you to take on still
• you still need to approach potential supervisors, who feel able to guide you through this project
• you also need to make this project your own, and determine some details as it is defined at a higher level, which would all go into your project proposal

• this is a design-heavy project (low code)
• the point of contact about this project is available to answer questions and confirm whether it is available for you to take on still
• you still need to approach potential supervisors, who feel able to guide you through this project
• you also need to make this project your own, and determine some details as it is defined at a higher level, which would all go into your project proposal

• the point of contact about this project is available to answer questions and confirm whether it is available for you to take on still
• you still need to approach potential supervisors, who feel able to guide you through this project
• you also need to make this project your own, and determine some details as it is defined at a higher level, which would all go into your project proposal

This is a project that is defined by colleagues from Bournemouth University Sustainability Team. The digital platform should create a supportive environment, enhance knowledge, encourage the use of efficient energy sources through tracking and monitoring of current uses and to account for climate responsibility by students and staff at BU. The goal is to create a Dashboard for members of the BU community who will be able to see current usage for various energy types, and their potential emissions contributions. Information to be displayed in a fun and creative manner which engages the target audience. The platform will support the work of BU’ Sustainability Team as they aim to achieve environmental targets.

This project is open for multiple students, though you need to work independently (cannot collaborate on literature review, methodology, development, etc).

• Problem: Inputting emoji through the visual keyboard is not great.
• Solution: Develop a few different ways to input emoji with custom swipe (or others) gestures.
• Evaluation: Comparative evaluation to see which of your new designs is best

• Problem: Accessibility on the web is still poor, there are many evaluation tools, yet they are cumbersome to use and their results can be difficult to interpret.
• Solution: Develop a new tool that improves displaying of WCAG errors to developers.
• Evaluation: Compare your tool against common tools such as WAVE, AXE, or Lighthouse

• NFTs (Non-fungible tokens) have been very popular since 2021 (e.g., https://www.nbatopshot.com/ or https://www.cryptokitties.co/). They have numerous advantages over common collecting, but how could we improve it or move it in new directions? This project could explore either new ideas for NFTs (I have some speculative ideas) or discover how to explain NFTs to users, other ideas could explore how to display, curate, and store NFTs.
  Tracking Cryptocurrency portfolios is often limited to the wallet/exchange or you need to export your wallet data to a service like Koinly (https://koinly.io/). The features are often limited. Could there be a better way for people to track their portfolios without needing to connect their wallets?
  
  
• Cryptocurrency and NFTs are potentially contributing to the climate emergency, yet are likely not going away anytime soon. Could you develop a system where you export someone's crypto portfolio and demonstrate how much it has cost the climate? Similar to http://cryptoart.wtf/ but for full portfolios, could you then allow people to offset this?

• NFTs is still seen as a solution looking for a problem. Yet many applications are yet to be seen realised. Consider using NFTs to track online course progression (e.g., Coursera).

• Problem: Accessibility is still an afterthought when producing content.
• Solution: Investigate ways of providing nudge support (https://dl.acm.org/doi/10.1145/3290605.3300733) during the production of content.
• Evaluation: Compare against typical ways of providing access feedback.

Problem: Media consumption could increase engagement with use of IoT Devices. Many houses now have IoT devices (e.g., Smart Speakers, Smart Lights etc). Could they be used or activated using media to improve engagement and audience experiences? Imagine your IoT lights changing colour when Darth Vader is on screen or even when his name is spoken. Or the lights flickering when Voldemorts name is spoken.

Solution: Using triggers within media, activate smart devices based on these triggers during media playback.

Evaluation: Explore audience reaction to system to evaluate engagement levels.

Various lung audio sounds (labelled and unlabelled) need to be investigated in order to deploy ML methods for understanding the chronic respiratory diseases. The aim here is to deploy scientific experiments in order postulate your hypotheses and verify them through prediction bases algorithms for the classification of the diseases. The ranking of such algorithms in terms of their performances is a key finding.

Shoulder pain is an issue that affects a significant portion of the population, with an annual prevalence of up to 46.7% and a lifetime prevalence of up to 66.7% according to Artus, et al, 2017. It is commonly managed in primary care settings and ranks as the third most common reason for musculoskeletal consultations. A considerable number of adults, approximately 3%, seek medical advice annually for new shoulder pain.

In this project, a technical solution for measuring shoulder motion needs to be developed. The solution should be based on MediaPipe Pose (https://developers.google.com/mediapipe/solutions). The research questions:

1. What are the requirements for measuring shoulder motion for recovery and/or diagnosis?
2. What are factors influencing the measurement accuracy and how can this be improved?

Strong programming skills are required. 

This project will also receive supervision from Prof. Carol Clark of the Faculty of Health and Social Sciences and Mrs. Emily Middleton of University Hospitals Dorset.

Shoulder pain is an issue that affects a significant portion of the population, with an annual prevalence of up to 46.7% and a lifetime prevalence of up to 66.7% according to Artus, et al, 2017. It is commonly managed in primary care settings and ranks as the third most common reason for musculoskeletal consultations. A considerable number of adults, approximately 3%, seek medical advice annually for new shoulder pain.

The project needs to review potential technologies for support. 

1. measuring positions using static pictures
   • What are supported technical solutions and comparison of different solutions? 
   • What are the requirements for implementation of the solutions?
   • What is the feasibility of implementing the different solutions using static pictures?
2. measuring position based on shoulder motion respectably  
   • What are supported technical solutions and comparison of different solutions? such as MediaPipe Pose, OpenPost
   • What are the requirements for implementation of the solutions?
   • What is the feasibility of implementing the different solutions based on shoulder motion?

This project will also receive supervision from Prof. Carol Clark of the Faculty of Health and Social Sciences and Mrs. Emily Middleton of University Hospitals Dorset.

How can you measure the performance of the library best?

• What test data represents a normal use case?
• How can you determine optimization opportunities?
• What optimizations are (most) effective?

xmlutil is found here

You will need good mathematical background and programming skills as well as interest in quantum computing.

Scientific papers published by researchers and scientists make significant contributions to the field of science. These papers are typically regarded as objective and impartial; however, it's important to recognize that emotions also play a pivotal role in shaping their findings. Understanding the emotions and motivations of researchers can foster a more empathetic approach to scientific collaboration and the decision-making process.

1. Are emotions disseminated through CAiSE conferences? 
2. Can we discern the prevailing emotions and emotional patterns within the abstracts of papers presented at CAiSE conferences?
3. Which method, between Lexicon-based and Machine Learning approaches, would be most suitable for identifying emotions in CAiSE conference papers? Why?

The growth of wireless devices and sensors usage have give rise to interest on understanding the behaviour of mobile ad hoc networks (MANET) (e.g., in terms of performance and resilience). In such networks, nodes are mobile and have limited connectivity capability. Under such conditions, the network topology evolve over time. In this project, the student is to study the topological properties of such networks with different mobility models under different settings. 

Student will be given pointers to relevant literature at the beginning of the project. 

Student is required to design and implement a basic MANET simulator with different mobility models. 

Student can freely choose the tool/programming language for this project (python, C/C++, MATLAB, Octave, existing network simulators, etc.).

(THIS PROJECT IS NOT AVAILABLE CURRENTLY.)

Objective: To Identify a subgroup of patients with acute clavicle fractures who would benefit from early surgical intervention

Background: This projects represents an exciting collaboration between the Trauma & Orthopaedic team at University Hospitals Dorset (UHD) and Bournemouth University.

The clavicle is a commonly fractured bone with approximatley 400 patients presenting to A&E at UHD per year. Patients can be treated wither using a simple sling or with surgery, which involves fixing the bone with plates and screws. The long term results of surgery and non-operative treatment are the same as long as the fracture heals.

However 15% of fractures treated with a sling alone do not heal (non-union). Most of these require delayed surgery (which is more difficult) and this leads to patient dissatisfaction.

At the moment there is no accurate way to predict who are the patients likley to develop non-union and would have therefore benefited from early surgery.

Similar studies using ML to predict hip replacement failure suggest that there are relevant features on x-rays that are iinvisible to the human eye. This technology has, as yet, not been applied to clavicle fractures.

Using the UHD fracture database we aim to develop and train a model to assess whether ML can see early x-ray features that predict clavicle healing?

A database of serial x-ray images, patient demographics and patient outcomes will be provided as well as support to provide clinical context to the research project.

Real world networks are often time-varying, i.e., the network topology changes over time. However, in the literature, most studies simplify these networks as static and thus, ignoring the time ordering of contacts which can be important to understand the information diffusion process taking place in the network. 

In this project, 

• Student is expected to simulate information diffusion processes in temporal networks and gain insights into the behaviour of these processes under different conditions. 
• Student will be given pointers to relevant literature at the beginning of the project.
• Student will be given access to various appropriate datasets.
• Student can freely choose the tool/programming language for this project (python, C/C++, Matlab, Octave, etc.)

Interested student can refer to the following paper as a start to understand the context:

https://www.nature.com/articles/s41598-019-43029-5

This project is associated under the Flexible and Complex Network (FLEXNET) Research Group. 

Many real-world systems are interconnected (e.g., power grid and communication networks). Their inter-dependency could lead to what is called cascading failures in which failure from one network may affect the functioning of the other networks which in turn may affect the first network, causing a catastrophic degradation of the entire system. 

Interested student may refer to the following paper to get the context of this project:

https://ieeexplore.ieee.org/document/7497224

In this project, the student is expected to implement and simulate such cascading failures and study the robustness of inter-dependent networks. 

• Student will be given pointers to relevant literature at the beginning of the project.
• Student can freely choose the tool/programming language for this project (python, C/C++, MATLAB, Octave, existing network simulators, etc).

This project is associated under the Flexible and Complex Network (FLEXNET) Research Group. 

Multiple organisations might collaborate on the creation of a software product, but have different expectations around trust and information classification.  If the organisations adopted Dev[Sec]Ops then what would be the risks of such organisations forming part of a single CI/CD pipeline.

Creating a risk model and, if there isn’t a repeatable/pre-existing process for coming up with it, a tool-supported process for incorporating Dev[Sec]Ops into software development, to form a single CI/CD pipeline for software creation

NIST SP 800-37 is an approach for ensuring security embedded into the design and operation of new and existing systems from the outset and through life.  Responsibility for implementing this approach rests not with accreditors, but with the teams responsible for delivering and operating the systems. 

Because the availability of substantial security expertise on project teams might be limited, responsibility for many risk management tasks falls on those with expert domain knowledge, but comparatively limited knowledge of security.  In theory, this situation should not inhibit the approach; many of the ‘PREPARE’ steps don’t assume they are completed by security experts.  In practice, however, non-security experts would benefit from tool-support to guide the capture, management and analysis of collected risk data.  Such tool-support should also support the means of outputting risk data in a format that would help accreditors obtain assurance that a project team is properly following the approach.

Student would need to be willing to pick up NIST SP 800-37 and CAIRIS.  To support the later, some experience with SQL, Python and JavaScript / Vue.js would be useful.

The main aim of the project is to learn about the difficulties of implementing a system-level fuzzer, as well as the benefits that it may provide. A secondary (significantly more complex) aim would be to have a way of identifying scenarios that result in radically different system behaviour for two different versions of the software under test.

The above figure roughly indicates the sort of prototype that a student might produce.

In today's software development landscape, Scrum has become the leading methodology for managing projects. However, there is still limited understanding of the complex dynamics within Scrum teams, particularly how personality traits influence their effectiveness. This study aims to address this knowledge gap by exploring the relationship between team members' personality traits and Scrum team effectiveness. The research will employ a quantitative approach to analyse and measure these impacts. 

This project focuses on developing an AI tool that translates complex clinical letters into simplified versions understandable by patients aged 12 to 16. The system will aim to ensure that young epilepsy patients, who often struggle with medical jargon, can comprehend their diagnosis and treatment plans clearly, improving their understanding and adherence to treatment.

Methodology:

1. Gather a dataset of complex clinical letters and simplified versions.
2. Train the AI to translate clinical jargon into language suitable for adolescents.
3. Test the tool's effectiveness on 20 real examples by comparing AI outputs with feedback from patients and healthcare providers.
4. Refine the translation model based on user feedback.

Expected Outcome:
Improved patient understanding of their medical conditions and treatment, leading to better compliance with medical advice and enhanced communication between doctors and younger patients.

The project investigates how cascading failure caused by overload spreading in a network affects the functioning of a network. The project takes the Motter and Lai model (see https://journals.aps.org/pre/abstract/10.1103/PhysRevE.66.065102) as the starting point to study how different networks suffers from such cascading failures. 

The project involves the implementation of the model and conducting simulation study for different networks and operating conditions.

This project is associated under the Flexible and Complex Network (FLEXNET) Research Group. 

AI adoption in finance is growing rapidly, but success depends not only on technology, but also on how professionals perceive and trust it. Many studies focus on technical systems or customer adoption, while professional readiness is less understood.

• Do finance professionals trust AI-driven decisions?

• What ethical concerns shape adoption?

• Can perceptions and behaviours be predicted using data-driven models?

Students will investigate these questions by analysing survey responses and applying statistical and machine learning techniques.

• Suitable for: All students but particular to Business Computing with Analytics students, Data Science students, AI students (final year or master’s).

• Skills developed:

  • Survey data handling and cleaning.

  • Statistical modelling (regression, multilevel models, SEM).

  • Machine learning classification and prediction (e.g., Random Forest, XGBoost, Neural Nets).

  • Model interpretation (SHAP, feature importance).

  • Academic writing and visualisation of findings.

• Data: A structured dataset collected from finance professionals across multiple countries.

• Deliverables:

  • A predictive model of AI adoption readiness.

  • A report discussing key drivers of perceptions, behaviours, and global differences.

  • Data visualisations and dashboards.

Problem Definition
Many expectant mothers face barriers to accessing reliable information on pregnancy, antenatal care, and postnatal health. Traditional healthcare systems are often resource-constrained, leading to gaps in education and support.

Approach

• Develop an AI-driven conversational agent (chatbot) trained on validated maternity and perinatal care guidelines.

• Tailor information delivery to trimester-specific needs, language preferences, and literacy levels.

• Integrate machine learning to identify when users may require additional emotional support or referral to a healthcare professional.

Expected Outcomes

• Prototype of a maternity-focused AI chatbot.

• Evaluation of acceptability, usability and accuracy.

You will need both software development and machine learning skills.

Problem Definition
Pregnancy outcomes are influenced by a combination of medical, social, and behavioural factors. Current care models often take a standardised approach, missing opportunities for personalisation.

Approach

• Build predictive models using demographic, lifestyle, and clinical data (e.g., BMI, smoking status, previous pregnancies, blood test results).

• Explore clustering methods to stratify women into different risk groups for complications such as gestational diabetes or preterm birth.

• Propose personalised care plans (frequency of visits, additional monitoring, dietary interventions).

Expected Outcomes

• Developing risk model for pregnancy outcomes and performing classification via machine learning.

• Framework for personalised maternity care supported by AI.

Shift-based emergency service workers face high stress, irregular hours, and operational pressures that negatively impact their physical and mental wellbeing. Existing wellbeing interventions often fail to accommodate the unique needs of shift workers, and there is a lack of digital tools that integrate operational demands with personalized health support. There is a need for systems that can deliver timely, context-sensitive wellbeing interventions aligned with workers’ schedules.

The project involves co-designing, developing,  usability testing, and long-term evaluation of the app functionalities. A key component of the project is developing the app to recommend wellbeing actions that are tailored to users’ schedules, ensuring notifications and prompts are delivered at optimal times for engagement and effectiveness. The shift management feature is central to the efficacy of the app, enabling personalized wellbeing support based on each user’s schedule. Additionally, the project will involve developing a framework to guide the recommendation and development of app features that address different health and wellbeing challenges for shift-based workers. 

Problem Definition
AI-driven mental health tools (such as chatbots, mood trackers, or AI-based support systems) are increasingly used to provide accessible, low-cost psychological support. While these systems can offer early interventions, they also raise critical ethical concerns and risks. These include privacy violations, algorithmic bias, over-reliance on non-human support, and unclear accountability when harm occurs.

Research Aim
To investigate the ethical challenges associated with AI-based mental health support systems, assess their potential risks, and propose guidelines or design principles for responsible deployment.

Approach

• Survey existing AI mental health support tools and ethical frameworks in digital health and AI.

• Identify risks such as data security, patient consent, algorithmic bias, misinformation, dependency, and lack of empathy.

• Propose an ethical evaluation framework for AI in mental health care.

• Evaluate scenarios where reliance on AI could delay professional help or exacerbate conditions.

• Suggest practical design guidelines to mitigate identified risks.

Expected Outcomes

• A structured analysis of the ethical risks and challenges of AI mental health support.

• A framework that can serve as a reference for evaluating future AI mental health technologies.

• Recommendations for on how to responsibly design and implement such systems.

Requires experimental study design and quantitative research skills.

Problem Definition
Elderly patients are often prescribed multiple medications (polypharmacy), which increases the risk of adverse drug interactions, missed doses, or overdosing. Traditional pill-box reminders do not adapt to changing medication schedules or risk profiles.

Approach

• Develop an AI-powered reminder system that personalises medication schedules and adapts to individual adherence patterns.

• Use computer vision (e.g., pill recognition via smartphone cameras) or IoT based pill dispensers to confirm adherence.

• Integrate AI-driven drug–drug interaction alerts.

Expected Outcomes

• A smart medication adherence prototype with risk alerts.

• Evaluation of usability and reliability for elderly users.

Requires good programming skills.

Problem Definition
In CNC-based manufacturing, the total production runtime is often extended not only by machine cycle times but also by human-in-the-loop delays. These include waiting for the next operator, tool changes, material handling, setup misalignments, or delayed responses to machine notifications. Such inefficiencies increase costs, reduce throughput, and hinder delivery schedules.

Research Aim
To develop an AI-assisted system for monitoring and optimising CNC process flows, with a focus on reducing runtime losses caused by delays between process steps and late operator responses.

Approach

• Survey the problem domain and collect/generate data on machine runtimes, idle times, alarms, operator interventions, and handovers between process steps.
• Use simulation and/or machine learning techniques to identify bottlenecks and patterns in where and why delays occur.
• Develop an AI model (e.g., predictive analytics or reinforcement learning) to forecast potential bottlenecks in real time. For example, predicting when an operator is likely to respond late to a machine prompt, or when tool changes are likely to exceed the standard time.
• Build a prototype dashboard that provides real-time alerts for upcoming delays.
• Compare runtimes, idle times, and operator response times before and after AI-based optimisation by using simulation.

Expected Outcomes

• Identification of the main causes of runtime delays in CNC workflows.

• A prototype AI-based monitoring and optimisation tool.

• Using simulation to evaluate the proposed solution.

Requires good programming and machine learning skills.

The EU Cyber Resilience Act 2024 (https://eur-lex.europa.eu/eli/reg/2024/2847/oj/eng)  is a new regulation that sets mandatory cybersecurity requirements for both software and hardware with a digital component that is sold in the EU market to meet new, much stricter design and implementation guidelines. The intention behind the act is to ensure products and services sold are better able to resist cyber-attacks, manage vulnerabilities, provide security updates, and report incidents, ultimately enhancing product security and empowering consumers to make informed choices. Products will carry the CE mark to signify compliance. 

The Data Use and Access Act 2025 (DUUA) is a newly enacted UK law that modernises regulations around digital information.
It updates data protection rules to encourage innovation and economic growth, while continuing to safeguard individuals and their rights.

Rather than imposing strict new requirements, most of the changes offer flexibility—giving organisations the chance to approach data use differently.

The intention is the changes will be introduced gradually between June 2025 and June 2026 making this a very timely project

The project idea is to create a guidance framework to help either a client or a client group (SMEs or charities) to implement the changes brought in by DUUA within the organisation. These changes include:

• New "recognised legitimate interest" as a lawful basis for processing personal information
• Introducing an "Assumption of Compatibility" that will allow some reuse of personal information in certain circumstances even if you don't have explicit consent 
• A soft 'opt-in' option for charities
• Clearer guidance on subject access requests and how to deal with complaints

The project will require you to find and work with one or more clients for both developing and evaluating the solution created

• the point of contact about this project is available to answer questions and confirm whether it is available for you to take on still
• you still need to approach potential supervisors, who feel able to guide you through this project
• you also need to make this project your own, and determine some details as it is defined at a higher level, which would all go into your project proposal

This project is open for multiple students, though you need to work independently (cannot collaborate on literature review, methodology, development, etc).

• the point of contact about this project is available to answer questions and confirm whether it is available for you to take on still
• you still need to approach potential supervisors, who feel able to guide you through this project
• you also need to make this project your own, and determine some details as it is defined at a higher level, which would all go into your project proposal

The increasing importance of the information domain is widely recognised, but this also means there is an increasing threat surface to be protected. Information can be transferred in different forms across multiple systems and across domains, such as cyber space or the electromagnetic spectrum (focussing on radio signals here). As information flows across different systems it creates correlations that form normal patterns of behaviour.

To understand threats to our systems we need to develop a pattern of life (PoL) that captures behaviours to understand what is normal, and then to identify anomalous behaviour that indicates possible threats. Scope here covers behaviours in cyber space and electromagnetic (radio) signals.

Understanding PoL and using it to detect threats will require; information models to describe PoL, data capture and analysis to describe a normal pattern of life and then processing to detect anomalous behaviours. Of particular interest here is PoL involving more than one domain (cyber/electromagnetic).

The project will review existing PoL models (including in the physical domain) and create an information schema and model to describe PoL across domains. This will include cyber and radio environments, and their interaction. It will then capture different methods for the processing and fusion to create a PoL description, and carry out anomaly detection. A software architecture will then be proposed for a complete threat detection system based on PoL.

Outcomes (Artefact)

·       Information model for CEMA PoL.

·       Software architecture for PoL based CEMA threat detection.

·       Recommendations for development roadmap.

Increasing application of agile/cognitive technologies, underpinned by AI/machine learning, means the environment is rapidly changing and the original programming of systems is no longer adequate. Wide availability of software defined capabilities, such as software defined radios, is a key enabler to allow reprogramming to match changing demands and environment. However, it is undesirable to have to return systems to a service centre for reprogramming, ideally this is done remotely with the system in-situ.

The mechanism to carry out remote reprogramming should not depend on system specifics and instead should be system agnostic. However, reprogramming creates a new attack vector and so security of the architectures needs to be considered. This will require common architectures in software and hardware, as well as an interoperable signalling system to transfer code and assure reprogramming. For example, you don’t want to ‘brick’ the device.

Review existing standards for reprogramming of systems and develop a set of desired requirements for a reliable and secure remote reprogramming capability. Identify alternative hardware/software architectures and assess them against the requirements. Provide recommendation on architectures and issues to be resolved requiring further research.

Outcomes

·       Documented requirements for assured remote reprogramming.

·       Assessment of alternative architectures.

·       A recommended architecture for assured remote reprogramming.

A Service Oriented Architecture (SOA) is a software pattern that provides flexibility to rapidly create and manage different workflows, drawing from a number of services. But where there are many available services, the ability to create the workflow needs to consider the user interaction and ease of use.

This project will propose an approach to designing user interfaces for SOA to make workflow creation and management fast and easy.

The project will first review applications of SOA to understand how services are defined and the need to create complex workflows. It will then define a set of requirements for a software application that helps a user build and subsequently maintain a workflow. It will finally, propose an approach to implementing the workflow tool. Exemplar pictorial illustrations of the interface will demonstrate ease of use through user stories.

Outcomes

• A set of requirements and proposed technical approach for a workflow design and management tool.
• Illustrations of different user stories.

In the context of security requirements engineering, Threat Modelling is undertaken during the early stages of the design process to identify potential areas of exploitation, leading to the specification of threats, vulnerabilities, and risks - and later, mitigating requirements.

Threat modelling typically uses a Dataflow Diagram (DFD) to identify areas of concern towards the processing, storage, and transmission/transportation of data, etc. STRIDE (or others) may also be applied, alongside attack trees (or obstacle models), helping towards identifying certain threat categories and potential weaknesses/vulnerabilities.

However, for non-security experts and occasionally students learning how to apply threat modelling with DFDs, there can sometimes be confusion about how to apply and interpret DFDs. Consequently, where there may be gaps in understanding and application, there may also be gaps in related analysis, thus potentially overlooking other important concerns (threats, vulnerabilities, and risks), which would go unmitigated.

There is, therefore, and need to explore how applying the process of Threat Modelling could be improved for non-security experts, students, and other applicable stakeholders. This could, for example, lead to a better approach towards explaining how to apply and interpret DFDs when threat modelling. It could also lead to a new approach, perhaps introducing a different model type for a more effective and understandable threat modelling process. The process would need to be applied by non-security experts, or complement existing security stakeholders’ expertise, providing a robust threat modelling approach towards identifying related threats and vulnerabilities, leading to the specification of risks and applicable requirements. A longer-term aspect of this output would be to incorporate the approach within a tool-support application, and which includes automation for elements of the process.

The type of expected artefact would therefore be dependent upon the chosen direction and focus of a related project.

Cellebrite’s / MSAB software is used to download and review data from recovered suspect and victim devices. Once recovered, the data provided to officers/CPS is sometimes not easily digestible within their current formats (PDF/Excel), e.g content, time stamp, location data and communications with other parties. The data has to be interpreted, and the key evidence located manually. As a result, large data sets are not easily presented to CPS. Often there are thousands of pages within produced reports, when all that may be relevant/required is only a fraction of what’s included.

Full data downloads sometimes hold previously connected SSID’s this matched with GPS would greatly assist Investigations, as well as a number of devices submitted on a job and the communication between those devices.

All phones are different with different data outputs.

Developing a software prototype that can work with/ingest the produced data from the electronic reports (Excel and/or PDF), generated by Cellebrite/ MSAB, and then output a more streamlined/condensed and digestible reviewing document/format that would provide instant feedback within review. For example, an infographic presentation of phone activity within a time period (Input by the user), that could be filtered to particular to applications (Input by the user) for ease of interpretation. The functionality could be expanded upon to help identify data relevant to the investigation as digital policing requirements evolve.

Any programming language/method can be used as per students preference – the ideal method would be to add the original file to digest the data but not to keep the data as this could be sensitive and personal. Likewise the output does not and should not be stored in the software. Ideally, it is downloaded when produced with no copy of data being stored within the software.

Risk appetites communicate how much an organisation, project, or system owner is able to tolerate risk. However, risk appetites can evolve in line with changes to the expectations of problem owners, and changes to the context the system operates in. Furthermore, it is not always clear how such changes are reflected when managing risks. 

This aim of this project is to identify and evaluate solutions for cohering risk appetite statements with risk models, such that the impact of changes to the former can be explored in the latter.