In this research project, we aim to develop a mmWave communications system based on reconfigurable metasurface. By adopting reconfigurable metasurface technology to ultra high frequency range communication technology, we can overcome inevitable limitations caused by using high frequency electromagnetic wave such as tremendous amount of path loss and shaded area due to the high straight forwardness characteristic.

The whole system is made up of phased array antenna transmitter, RF-DC convert circuits receiver and 1-bit coding metasurface reflector. To tackle its low power transfer efficiency between transmitter and receiver, which is the foremost obstacle of radio frequency (RF) based WPT, we have implemented high-efficiency microwave beam synthesizing with multiple antennas. Inevitable shaded area due to the various kinds of obstacles between transmitter and receiver has been solved by deploying reconfigurable metasurface near the shaded area. In the case that transmitter detects obstacle in the middle, energy beam from the transmitter would be steered to metasurface and then by converting the phase of incident electro-magnetic waves of each cell in metasurface, transmitted power can be focused to receiver as well. With proposed convergent power transfer algorithm that handles all hardware components at the same time, optimized power transfer efficiency can be achieved regardless of whether the obstacle detected or not.

In a communication system utilizing beamforming, a beam alignment process is essential, and the overhead in this process is significant. In the case of A6G (Above 6GHz), more than 1000 beams are used. In order to satisfy various requirements of 5G, efficient beam alignment method is required in each scenario, the beam must be electrically steering In the direction of the terminal belonging to the cell, interference should be minimized through nulling. That is, a beamforming technique is needed to quickly find a terminal device where it may be located, start communication by aligning the beams, track movement well, and quickly recover when a communication link is broken due to surrounding conditions. We have optimized the algorithm using the Gradient Descent method so that it has various directivity and nulls and a radiation pattern of various beam-widths.

We propose a method of adjusting only the phase of the feed signal by optimizing it with the objective function and computation process. We propose a beam synthesis method that can be effectively used in a communication system with a lot of limitations, and propose an initial access method of a millimeter wave communication system utilizing a planar array antenna using a synthesized beam.

In order to effectively utilize the beam, research for forming various beams through beam synthesis is required. And initial access process is introduced by using several beam form to shoot the optimum beam.

Today, as the use of communication devices increases, a high-density network environment has also increased, and thus limited channel resources must be efficiently used. Since the existing fixed algorithm was developed to be used in the overall communication environment, there is a limit to the performance improvement over a certain level in the actual environment. If a communication algorithm using a reinforcement learning model is used, it is possible to maximize the use of channel resources by learning the data of the current environment, and to find a communication algorithm optimized by itself according to the changing environment.

Therefore, in this study, we implement a frequency sharing environment simulator and develop a reinforcement learning model that optimizes communication algorithms according to the changing environment using the simulator.

The ultra-dense network (UDN) is a next generation communication technology that can maximize spectral efficiency per unit area by simultaneously transmitting and receiving information through the same spectrum to multiple terminals through cooperative communication between a plurality of base stations arranged at very high density. In this research, we design UDN test bed based on software defined radio (SDR) platform to improve spectral efficiency of UDN structure in ultra high density environment and verify UDN core technology.

We can modify the parameters such as modulation order, transmit power, and basis vector size, etc. in the top-level VI of the LabVIEW code and check the performance metrics such as bit-error rate (BER), signal-to-noise ratio (SNR), channel gain, and signal constellation of each receiver. We have conducted experiments that measure these performance metrics according to various parameters.

And one of the many challenges in the development and evaluation of these wireless systems is testing and verification. Therefore, the radio system must be tested to ensure specified requirement. But field tests are expensive, time-consuming, and difficult to repeat. Simulation can overcome these problems, but ultimately it is limited to accuracy or excessive running time. One way to compensate for these limitations is the real-time HIL channel emulation. This can compensate for the gap between simulation and field testing. A good channel emulator can test a wireless system in repeatability, high accuracy, and various network scenarios.

Cell-Free MIMO is a wireless communication technology that provides simultaneous service to multiple users across a wide area, without clearly defined cell service boundaries. In contrast to conventional cellular communication, it uses a user-centric cell with protected UEs. This technology allows a cluster of nearby APs to be assigned to each UE, with multi-antenna APs and UEs assumed.

The main challenge of Cell-Free MIMO technology is to maintain a set of “protected UEs” for each cell and minimize interference to them. The system also requires real-time exchange of information and calculations between base stations, which presents significant computational and architectural challenges.

Cell-Free MIMO technology is designed for distributed CPU, which means there is no dedicated hardware for CPU. Instead, each AP retains the capability of CPU so that all APs have an identical architecture. Each cell has a primary AP and others are called secondary APs. The primary AP performs high PHY signal processing and distributes the data stream to secondary APs. It also gathers channel information and computes the precoding matrix. This technology assumes multi-layer communications, and downlink OFDM communications are used.

Through-Wall MIMO Radar is a technology that allows for the collection of information about objects and people located in the space behind a wall or obstruction that blocks the line of sight. By using MIMO technology, which utilizes multiple antennas to simultaneously transmit and receive signals, a more accurate image of the target can be generated. The radar measures time delay and phase shift of reflected signals to determine the location, size, and movement of the object or person being detected.

In this research, an IR-UWB MIMO Radar system is configured with Multi-Input, Multi-Output (MIMO) technology on the Impulse Radar Ultra-Wide Band (IR-UWB) sensor. A large amount of RF signal data is collected for many people and various postures.

The research focuses on RF signal-based pose estimation deep learning model, which applies collected RF signal data to a feature extraction module. The posture of a person located in an invisible space is estimated using Transformer, an attention-based deep learning model. This research can have significant implications in fields such as surveillance, rescue operations, and security.

The rapid growth of wireless communication technology has led to an increasing demand for efficient utilization of wireless resources. Our research focuses on developing artificial intelligence-based spectrum access technology that can optimize WiFi resource utilization efficiency while protecting existing wireless stations in the 6 GHz.

To achieve this goal, we analyze the 802.11ax network technology and the Distributed Coordination Function (DCF) transmission method in existing Media Access Control (MAC). We also investigate the technologies and requirements such as Orthogonal Frequency Division Multiple Access (OFDMA), channel bonding, preamble puncturing, and Multi-User Multiple Input Multiple Output (MU-MIMO).

We have developed an 802.11ax based communication simulator and agent that can be implemented in the form of Discrete Event Simulation (DES). We have created a user-friendly interface using Python and visualization tools to aid in the simulation process.Moreover, we have created reinforcement learning models using value-based algorithms such as Deep Q-Network (DQN), Double DQN (DDQN), and Rainbow DQN, as well as policy-based and actor-critic algorithms such as REINFORCE, Advantage Actor-Critic (A2C), and Proximal Policy Optimization (PPO).

Capacitive touch screen technology has become an integral part of modern digital devices such as smartphones, tablets, and laptops. The ability to sense the presence and location of a touch through a change in capacitance has made capacitive touch screens a preferred choice over resistive and other types of touch screens.

Our research focuses on the implementation and improvement of touch screen panels using capacitive touch screen technology. We aim to develop a high-performance touch screen panel that can accurately detect multiple touch points and provide a smooth and responsive touch experience.

To achieve this goal, we have developed a technique that includes Tx and Rx electrodes to detect the touched points by measuring the capacitance changes. We also represent the touched basic cell with mutual capacitance for simulation purposes.Furthermore, we have developed a software-defined touch display platform that includes 20 Tx electrodes and 40 Rx electrodes. We utilize PXI system-linked DAC and DAC modules to control the mutual capacitance of multiple cells, allowing us to implement multi-touch situations. We have combined the implemented touch screen panel with the PXI system, providing scalability to operate on FPGA and CPU.