– ICON LAB
ICON LAB: We apply machine learning techniques and neural networks to various communication systems. We have developed a high-performance Wi-Fi system with Spiking Neural Networks (SNNs), a trajectory prediction system for self-driving cars, and a new radio signal receiving system that can avoid interference in radars.
– Spiking Neural Networks (SNNs)
SNNs are next-generation neural networks using spikes for delivering information. They show high accuracy while consuming less amount of energy than other neural networks. Because of this characteristic, SNNs can dramatically increase the communication performance of portable devices when they are implemented in portable devices. Currently, we apply SNNs to various communication protocols and improve the learning rules of SNNs.
– Autonomous driving
We develop key primitives for realizing an autonomous driving system using V2X and machine learning. In particular, we focus on predicting the future trajectory of neighbors, which is imperative to the safe driving of self-driving vehicles. To this end, vehicles share their kinematic data (e.g., position, velocity, acceleration, yaw rate) via periodic beaconing and utilize machine learning techniques to predict the trajectories of their neighbors.
– AI-inspired wireless communications and networks
Our goal in this topic is to design intelligent algorithms that improve wireless communications and network systems. For this purpose, we leverage a model-based approach (mathematical optimization) as well as data-driving approach (machine learning). To demonstrate the validity of our algorithms, we evaluate performance using several techniques: mathematical models, network simulations, and sa oftware-defined radio platform (e.g., USRP).