Fatima, Haneen (2026) Internet of mirrors for personalised services: from conceptual vision to intelligent orchestration of a distributed IoT ecosystem. PhD thesis, University of Glasgow.
Full text available as:|
PDF
Download (14MB) |
Abstract
Smart mirrors have emerged as interactive platforms for personalised health, beauty, and assistive services, but existing systems predominantly operate as standalone or application-specific devices. This thesis introduces the Internet of Mirrors (IoM) as an interconnected Internet of Things (IoT) ecosystem in which smart mirrors, users, professionals, service providers, and distributed computational resources operate collectively. Its novelty lies at the ecosystem level, providing a unified framework for coordinating heterogeneous mirror-based services, stakeholders, and computational resources.
The research first defines the conceptual and architectural foundations of the IoM through a stakeholder-oriented Consumer, Professional, and Hub hierarchy and corresponding functional layers. Practical feasibility is demonstrated through an automated dental smile-analysis application integrated into a smart-mirror interface, achieving 81.6% classification accuracy for excessive gingival-display detection.
The thesis then evaluates IoM application execution across heterogeneous computational infrastructure. Edgeâcloud experiments and a physical multi-tier testbed show that performance depends jointly on processing capability, communication overhead, network conditions, workload characteristics, and concurrent demand. Consumer-only execution produced end-to-end latency of approximately 1.5 s, while higher-tier offloading reduced this to approximately 0.8 s and lowered Consumer-tier resource utilisation.
To support increasingly diverse workloads, a Context-Aware Task Distribution (CATD) framework is developed using task requirements, computational capability, network quality, geographic proximity, and data sensitivity. Across bandwidth degradation, node failure, and increased workload conditions, CATD maintained higher taskcompletion rates than the evaluated single-factor baselines, demonstrating the value of integrated contextual assessment.
Finally, the thesis extends coordination beyond the IoM through a reinforcementlearning-based orchestration mechanism for interaction with the Internet of Sounds (IoS). The formulation incorporates ecosystem origin, modality, hardware compatibility, system state, and application context to coordinate visual and audio workloads across shared resources. Evaluation shows lower p95 latency and higher deadline satisfaction than cloud offloading for Live-Interactive tasks, while action masking and task deferral maintain feasibility under modality-dependent resource scarcity.
Collectively, the thesis progresses the IoM from a conceptual smart-mirror ecosystem to an experimentally and computationally evaluated distributed architecture.
| Item Type: | Thesis (PhD) |
|---|---|
| Qualification Level: | Doctoral |
| Subjects: | T Technology > T Technology (General) |
| Colleges/Schools: | College of Science and Engineering > School of Engineering |
| Supervisor's Name: | Taha, Dr. Ahmad, Mohjazi, Dr. Lina and Imran, Professor Muhammad Ali |
| Date of Award: | 2026 |
| Depositing User: | Theses Team |
| Unique ID: | glathesis:2026-86340 |
| Copyright: | Copyright of this thesis is held by the author. |
| Date Deposited: | 11 Nov 2026 15:58 |
| Last Modified: | 12 Nov 2026 01:06 |
| Thesis DOI: | 10.5525/gla.thesis.86340 |
| URI: | https://theses.gla.ac.uk/id/eprint/86340 |
| Related URLs: |
Actions (login required)
![]() |
View Item |
Downloads
Downloads per month over past year

Tools
Tools