Dynamic edge artificial intelligence models to empower IoT-based systems

Project summary

• Develop an artificial intelligence (AI) model and data problem description ontology to catalogue AI techniques and develop a dynamic AI selection/ranking algorithm, which allow IoT (Internet of Things) systems to automatically apply the best solution for a task in dynamically changing environments
• Create a dynamic AI edge framework that deploys suitable and proven AI models at runtime close to the data generated from IoT-based systems. The framework will increase computational efficiency, reduce the latency of responses and reduce security and privacy risks
• Evaluate the framework and algorithms in real-world IoT-based applications, initially in a health care domain (such as intelligent non-invasive blood property probes, or pain detection based on skin discolouration), but extending to other areas such as predictive maintenance

The demand for real-time, affordable, and efficient intelligent IoT system is increasing exponentially due to the fast development and revolution of AI and mobile sensor technologies. The proposed project aims to efficiently bridge the gap of invoking desired AI models to support real-time IoT system according to the dynamic context needs. Currently, IoT systems only use AI models embedded in specific applications or connecting to cloud AI services. However, there are three major limitations:

1. The IoT system is designed to communicate with only one AI model and respond to a single type of task in a static style. The system requires re-engineering to adapt for a dynamic environment and changing context, which are encountered by IoT systems
2. The cloud-based framework contains high risks of latent response, data transmission privacy and security. In many application domains, these risks are the fatal factors contributing to system failure and lack of user uptake
3. There is currently no solution enabling automatic identification and selection of the most suitable AI model by considering the features of the existing AI models, datasets used for the training, and transformation capabilities

Keeping these facts in mind, edge computing technology, semantic data and transform learning together can provide a suitable solution. The research will provide scientific novelties on:

1. Modelling and cataloguing high-quality AI models with descriptions on all aspects of the features for different domains. This can create the first AI model metadata repository with registration function available to researchers and application development communities as a service
2. A runtime edge configuration algorithm will provide efficient and flexible AI models to IoT-based system or application dynamically adjusting to context. Context-awareness is the unique highlight of the research, turning the IoT system from passive configuration to actively adapting to tasks by employing the best AI models

Research centre

Data Science Research Centre 

Entry requirements

Applicants will need either a first-class or upper-second (2:1) honours degree and preferably a masters degree from a UK university in a relevant subject such as Computer Science, Engineering, Mathematics, or qualifications that we consider to be equivalent.

International students may also need to meet our English language requirements. Find out more about our entry requirements for international students. 

Project specific requirements must align with the University’s standard requirements.

How to apply

Please contact Dr Hongqing Yu (Harry) (h.yu@derby.ac.uk) in the first instance for more information on how to apply.

The University has four starting points each year for MPhil/PhD programmes (September, January, March and June). Applications should be made at least three months before you would want to start your programme. Please note that, if you require a visa, additional time will be required.

Funding

Self-funded by student. There is a range of options that may be available to you to help you fund your PhD.

Supervisors