NT2203: Advanced Internet of Things Technology

Introduction

Welcome to NT2203

This course examines modern Internet of Things systems, with the emphasis placed on edge intelligence, TinyML and digital twin enabled architectures. We study how intelligent IoT applications are designed and deployed under real-world constraints on latency, memory, energy, bandwidth and privacy, moving beyond cloud-centric solutions. Students develop a practical system-level understanding through hands-on capstone work that delivers deployable edge intelligence, progressing from a connected prototype through on-device inference to a cognitive system that explains its own behaviour in natural language.

Course Information

Course Code

NT2203

Level

Graduate; academic year 2025-2026

Structure

12 weeks: 8 taught sessions and 4 capstone sprints

Core Focus

Edge intelligence, TinyML, digital twins and generative AI on device

Complexity Framework

Depth of Implementation

Every capstone is assessed against the depth it reaches. Level 1 establishes the baseline, Level 2 adds on-device intelligence, and Level 3 adds semantic reasoning over the twin.

Instructor

Dr. Dang Van Huynh

Lecturer, Department of Computer Networks, Faculty of Computer Networks and Communications

danghv@uit.edu.vn

Technology Stack

Devices and Modelling

Cortex-M boards DTDL / JSON-LD 3D twin viewers

Connectivity

MQTT / CoAP Matter / Thread LoRaWAN NTN (3GPP Rel 17/18)

Edge Intelligence

TensorFlow Lite Micro Quantisation Federated Learning K3s

Cognition and Trust

SLM / LLM APIs X.509 identity Secure boot Time-series / graph DB

Session Timeline

Scroll to view all 12 weeks

▼ Scroll

Select a Session

Tap a session from the list above to view details here.

Capstone Pathway

Project work runs through the semester and is defended in the four sprint weeks

Phase 1: Proposal and Design

Map team skills, model the telemetry with DTDL, then submit the project proposal and pass the architecture design review.

Weeks 1-8

Phase 2: Level 1 Hardware Sprint

Demonstrate robust physical links and a realistic power budget, with real sensor data transported over MQTT or CoAP.

Week 9: connectivity demonstration

Phase 3: Level 2 Intelligence Sprint

Demonstrate on-device inference or high-fidelity twin data, and report the accuracy, latency and memory footprint achieved.

Week 10: edge AI demonstration

Phase 4: Level 3 Cognitive Sprint

Demonstrate semantic value through generative AI integration, so that the system explains its own state in natural language.

Week 11: generative AI demonstration

Phase 5: Final Capstone Defence

Present the end-to-end scenario walkthrough and submit the final technical report, covering the design decisions and the constraints that shaped them.

Assessment

Target Skill Profile

Component	Evidence	Weight
A1. Final examination	Written assessment across the taught modules, from the modern IoT stack through to system architecture	50%
A2. Capstone project	Sprint demonstrations at Levels 1 to 3, final end-to-end walkthrough and technical report	35%
A3. Attendance and participation	Weekly sessions, design review and peer feedback during the sprint defences	15%

Level 1 Expectation

It Works

Real hardware acquires sensor data reliably
Telemetry is transported over MQTT or CoAP
Results are visualised and the power budget is reported

Level 2 Expectation

It Thinks

TinyML inference runs on the device itself
Anomaly detection or a high-fidelity twin is demonstrated
Accuracy, latency and memory footprint are measured

Level 3 Expectation

It Speaks

Generative AI is integrated for log analysis or advice
Insights are delivered through a natural language interface
The semantic value added over Level 2 is argued explicitly

NT2203 | Advanced Internet of Things Technology