SIH@SCH | Research

Research Area

We study how intelligent decision-making can improve wireless communications and IoT systems especially under dynamic channels, massive connectivity, and real-time constraints.

Focus

AI-Driven Wireless Communications for IoT & Edge Systems

Our research focuses on integrating Artificial Intelligence (AI) with Wireless Communications in IoT and edge environments. While modern AI models achieve high performance, they are often too heavy for deployment on resource-constrained devices due to high computational, memory, and energy requirements.

AI + Wireless

We aim to design lightweight, distributed AI-driven optimization frameworks that intelligently adapt network behavior while minimizing overhead. The goal is to optimize latency, energy consumption, and throughput in dynamic wireless environments, enabling practical and scalable IoT systems.

Example intuition

In a network of IoT devices, each node learns when to transmit, when to wait, and when to collaborate. Using Reinforcement Learning, devices adapt communication strategies such as power allocation and scheduling in real time. With Federated Learning, devices collaboratively improve models without sharing raw data, reducing communication overhead while preserving efficiency. This leads to fewer unnecessary transmissions, lower energy consumption, and more scalable wireless systems.

Reinforcement Learning Federated Learning Edge Intelligence Resource Optimization

Intelligent ultra-reliable and low-latency communication

We study ultra-reliable and low-latency communication (URLLC) from an AI-driven perspective, focusing on networked control systems in IoT and edge environments. These systems require timely and reliable delivery of control and sensing data under dynamic, uncertain, and resource-constrained wireless conditions.

AI + URLLC

Instead of relying solely on rigid protocol designs, we explore how lightweight intelligence can adapt communication behavior in real time. Our approach focuses on reducing signaling overhead, improving transmission efficiency, and enhancing robustness by enabling systems to make context-aware communication decisions under strict latency and reliability constraints.

Example intuition

In a cooperative autonomous driving scenario, vehicles exchange short control messages to coordinate movement. Instead of blindly transmitting at every time step, each vehicle uses learned policies to decide when to transmit, when to defer, and how to encode information efficiently. This reduces collisions, minimizes latency, and maintains reliability even in congested or unstable wireless environments.

Low-latency intelligence Adaptive communication Collision mitigation Edge-aware optimization

Data-driven optimization for wireless systems and IoT

We investigate data-driven and AI-based optimization for complex wireless systems, spanning both network-level intelligence and physical-layer components. As IoT and 5G/6G systems grow in scale and heterogeneity, traditional model-based approaches become less effective, motivating adaptive and learning-based solutions.

AI + Wireless Systems

Our work focuses on designing lightweight and efficient optimization strategies that improve system performance while respecting real-world constraints. This includes integrating learning mechanisms into communication processes, reducing control overhead, and enabling intelligent adaptation across both network protocols and hardware components.

Example intuition

In practical wireless systems, components such as cavity filters require precise tuning to maintain signal quality. Instead of manual or static configurations, we explore rule-based and Reinforcement Learning approaches that automatically adjust tuning parameters based on observed performance. The system learns how to optimize frequency response over time, reducing human intervention while improving efficiency and stability in dynamic environments.

Reinforcement Learning System optimization Hardware-aware AI Adaptive control

WiFi sensing and device-free perception

We explore WiFi-based sensing as a means of enabling device-free perception using wireless signals. By leveraging channel state information (CSI), wireless systems can capture subtle environmental changes without relying on cameras or wearable devices.

WiFi Sensing

Our research focuses on extracting meaningful patterns from noisy wireless signals using lightweight signal processing and AI techniques. The goal is to enable robust sensing capabilities such as presence detection, motion understanding, and environment awareness under practical constraints.

Example intuition

A WiFi transmitter and receivers are placed in a room. When a person moves, the wireless signal patterns change due to reflections and obstructions. By analyzing these variations, the system can detect presence and infer movement without any cameras or wearable sensors.

CSI analysis Device-free sensing Signal processing Edge intelligence