In this paper, we propose a 3D-IC PDN impedance prediction method using deep learning (DL). Unlike conventional application process (AP) and printed circuit board (PCB) design, our method proposes an effective solution to the problem of PDN design of 3D-IC, which requires accurate and fast impedance evaluation. For prediction, we construct regression neural network model based on VGG-16 with additional layers. Additionally, boundary element method (BEM) and Voronoi diagram are used to generate training data. We verified the proposed method in terms of accuracy and time. The impedance can be predicted nearly 4 times faster than ground truth which calculated by BEM with a low error.

Given the interest of researchers for wireless communications in electromagnetic (EM) shadow areas, a magnetic field communication has emerged to overcome RF challenges based on magnetic permeability characteristics. However, magnetic field communication has the demerit of a short communication range. Considering this limitation, we have proposed the GMI sensor-based field receiver to obtain immunity to low-frequency noise and high sensitivity for extension of communication range in previous studies. Further, this paper designs a magnetic field communication system with a GMI sensor-based receiver including a transmitter. Even though the basic concept of magnetic field communication is taken from previous studies, this study conducts the different realization of the system with transmitter system and communication link. Besides, this paper proposes the experiment of magnetic field communication using a GMI sensor-based system to verify its applicability in the practical environment for the first time. Bit error rate (BER), spurious-free dynamic range (SFDR), and random noise are measured with an on-off keying (OOK) communication link to analyze the stability and reliability of the system. The experimental verification and analysis entail the conditions of communication distance and channel medium to mimic practical communication in atmospheric, underwater, and underground environments. A comprehensive analysis of the system design and experimental verification for application in a practical environment show the possibility of the proposed system to realize improved wireless communications in EM shadow areas.

Radio reception relies on the medium which determines the propagation characteristics of the electromagnetic fields carrying the information. The permittivity varies greatly depending on the medium, but it remains nearly constant, except when magnetic materials are used. For this reason, magnetic fields, typically affected by permeability, can be utilized in microwave challenging environments. In this paper, a new approach based on the giant magnetoimpedance (GMI) effect is presented. The proposed GMI-based receiver has an effective double-superheterodyne topology, where “effective” means that the receiver actually has a single mixer but appears to have added a virtual mixer due to the GMI effect. The magnetic field-to-voltage conversion ratio (MVCR), the spurious free dynamic range (SFDR) and the receiver sensitivity are characterized, and from these results the optimal operating conditions of the fabricated receiver are obtained. Additionally, wireless digital communication using on-off keying (OOK) is demonstrated and transmitted and received waveforms are compared, with the final demodulation result of the receiver showing that the transmitted digital data are precisely extracted.

In this paper, a fast SerDes equalizer effect-included worst eye diagram estimation method is proposed for high-speed serial links. The proposed approach analytically integrates the effects of linear and nonlinear equalizers in the frequency and time domains, respectively, based on the behavior of the equalizers. This approach enables fast single-bit response (SBR) calculations without requiring input/output buffer information specifications, i.e., an algorithmic model interface (IBIS-AMI) model-based transient simulation. Additionally, a modified peak distortion analysis (PDA)-based worst eye estimation method was employed, including the outer contours. To validate the proposed method, tests were conducted in a peripheral component interconnect express (PCIe) Gen 6.0 channel environment configured with various equalizer combinations and different loss conditions. The proposed method achieves eye contour results comparable to those in conventional circuit simulations, operating up to 255 times faster than bit-by-bit methods and up to 66 times faster than statistical methods. The proposed method offers significant advantages in environments that require rapid exploration of optimal channel and equalizer combinations, providing substantial time savings and efficient evaluations.

In this paper, a deep reinforcement learning (DRL) approach that optimizes the power/ground ball map design within the ball grid array (BGA) package of 3D-ICs considering multiple-power-domain (MPD) environments is proposed. With the adoption of MPD in 3D-ICs, the BGA in the package substrate, which routes the power and ground to the appropriate power domain and components from the board, faces design challenges such as highly complex power noise between the domains and allocations considering multiple power densities. To address this design complexity, a U-net-based DRL method is adopted, which trains an optimal policy to obtain an optimal power/ground ball map to mitigate simultaneous switching noise (SSN). The U-net enables the conservation of semantic and spatial information simultaneously, allowing the proposed method to optimize the complex ball map effectively within the BGA package for MPD. Furthermore, the proposed method shows reusability and scalability by training with variously displaced current source center (CSC) balls and various ball map sizes while maintaining the power grouping design. The optimality performance of the proposed method is verified by comparing it to conventional optimization algorithms such as random search (RS) and a genetic algorithm (GA). The proposed method outperforms RS and GA in terms of the execution time and optimality performance across various tests with different ball map sizes and CSC positions.

In this paper, a tunable multi-split transmitting (TX) coil for a wireless power transfer (WPT) system that accommodates a wide range of distances between the TX and receiving (RX) coils is proposed. This method enables the WPT system to maintain a stable and consistent output power supply to the load, regardless of variations in coupling coefficients caused by changes in the distance between the TX and RX coils. The tunable multi-split TX coil can operate in various modes depending on how the wire connections between each TX coil are configured. This approach adjusts the inductance value of the TX coil for different conditions, using the same amount of coil as a conventional single-loop TX coil. The results show that by adjusting the TX coil to three different modes as the airgap varies from 50 mm to 250 mm, consistent output power is achieved with smaller variations in the input current and voltage compared to those in a conventional system. A conventional system requires an input voltage increase of approximately 529.64%, while the proposed system requires only a 42.93% increase. The proposed system enhances the power transfer capacity of the WPT system, particularly when operating in an over-coupling state. This approach provides a stable output power supply with a standardized and simplified TX coil structure.