Artificial Intelligence of Things

Aims

Artificial Intelligence of Things (AIoT) represents the convergence of intelligent data-driven models with interconnected physical systems. This module introduces both the foundational concepts of the Internet of Things (IoT) and the emerging shift towards embedding intelligence within these systems. It aims to develop an understanding of AIoT as a progression from traditional IoT towards closed-loop intelligent systems, emphasising system architecture, data-centric design, and the deployment of intelligence across heterogeneous and resource-constrained platforms.

By the end of this module, students will be able to: Understand the key components and architectures of AIoT systems across device, edge, cloud, and AI layers. Analyse trade-offs between latency, energy, communication, and model performance, and apply these in the design of AIoT systems under real-world constraints. Understand the role of distributed intelligence, including edge and on-device computing, in enabling real-time AIoT applications. Explore emerging AIoT paradigms, such as LLM-based agents and digital twins, and evaluate their practical challenges in real-world systems.

Syllabus

• Lecture 1: Introduction to AIoT
• Lecture 2 : Things 1: Sensors & Actuators Lecture
• Lecture 3: Things 2: Embedded Systems Lecture
• Lecture 4: Connectivity 1: Networking Fundamentals Lecture
• Lecture 5: Connectivity 2: Messaging Protocols Lecture
• Lecture 6: Sense Making 1: Data Processing
• Lecture 7: Sense Making 2: Cloud & IoT Platforms
• Practical 1: IoT Application Design Lecture
• Lecture 8: Energy Constraints and Solutions
• Lecture 9: Intelligence 1: Edge Computing Lecture
• Lecture 10: Intelligence 2: On-device Computing Lecture
• Lecture 11: AIoT with LLM Agents
• Lecture 12: Case Study: Smart Factory AIoT with Digital Twin
• Practical 2: Bringing Intelligence to IoT
• Lecture 13: Industry Guest Lecture
• Lecture 14: Academic Guest Lecture

Recommended reading

[1] Artificial Intelligence of Things (AIoT) https://www.sciencedirect.com/book/edited-volume/9780443264825/artificial-intelligence-of-things-aiot

[2] Internet of Things: Concepts and System Design https://link.springer.com/book/10.1007/978-3-030-41346-0

[3] Edge Intelligence: From Theory to Practice https://link.springer.com/book/10.1007/978-3-031-22155-2

Assessment - Part II Students

• The assessment consists of two assignments: Assignment 1 (40%) will be an in-class invigilated lab examination with Google Colab notebook, lasting for 1 hour. The student will be given a pre-trained model and a small dataset (downloadable from Colab), and the aim is to compress the model to a given size while maintaining as much as the accuracy. It assesses both the technical knowledge and implementation skills about on-device model inference. The students are expected to provide explanations for code snippets in different blocks to gain marks (in case the code is not executable). All online resources (e.g., tutorials, LLMs) are strictly not allowed and the invigilator will check around throughout the exam.
• Assignment 2 (60%) will be an in-class invigilated written examination, lasting for 1.5 hours. The students will be provided with a question paper and they need to type the answer in an online submission platform. It mainly assesses students’ understanding of AIoT knowledge covering different aspects. Two real-world IoT case studies will be provided in the question paper, and students are asked to answer multiple questions which form a complete design of the AIoT solution when combined. All online resources (e.g., tutorials, LLMs) are strictly not allowed and the invigilator will check around throughout the exam.

Self-certification for this module may not be permitted.  This matter will be considered at the meeting of the Faculty Board for Computer Science and Technology on Tuesday 14 October and the outcome will be communicated to students after the meeting.