Mark A. Beach

With the increasing difficulty of microwave circuit design for higher frequencies, there has been a surge of interest in artificial intelligence (AI)-aided electronic design due to its exceptional reasoning and problem-solving abilities. AI-aided electronic design is considered data-hungry because it relies on the quantity and quality of labeled data to train and optimize models. However, collecting a large amount of data is challenging for individual designers due to the high cost and computational burden. Furthermore, diverse and high-quality design data is distributed among multiple entities, and sharing raw data with high commercial value is unfavorable and could cause a breach of confidentiality. In this regard, federated learning (FL) can be introduced to tackle the dilemma between joint AI-aided electronic design and commercial confidentiality. However, unlike always-online environments, microwave software does not necessarily need to operate continuously and may have intermittent network connectivity. Additionally, various projects may have distinct specifications for circuit design. These factors pose challenges for implementing classical FL in microwave circuit design applications. The proposed RetroFL technique innovatively addresses these issues, tailoring FL specifically for microwave circuit design scenarios for the first time. By using this method, participants can train a generalized model asynchronously while safeguarding the secrecy of distributed data. Experimental results are provided to verify the effectiveness of the proposed retrospective FL method in modeling and designing microwave circuits. The proposed retrospective FL method allows the AI model to acquire knowledge from diverse sources, thereby enhancing the performance of AI-based circuit design algorithms.

This paper proposes a closed-form low-complexity method for RIS configuration in a multi-user cell-free massive MIMO (maMIMO) network. Leveraging the large-scale fading (LSF) distribution and spatial diversity of the distributed network architecture, the proposed method jointly enhances the end-to-end channel gain and the degrees of freedom for interference suppression. An LSF-based algorithm is introduced to maintain fairness among user equipment (UEs). We have analysed the proposed method under short-term and long-term RIS configuration schemes and different CSI acquisition strategies. A closed-form channel estimation error variance for $b$ -bit digital RIS is derived, and the non-negligible impact of digital RIS on pilot contamination is quantified. To explore different trade-offs between spectral efficiency (SE) maximisation and fairness, we also propose an element-selection based RIS configuration method. To verify the performance, we derive three achievable SE equations for perfect CSI and imperfect CSI scenarios with a fully centralised cell-free maMIMO operation. Numerical results show that the proposed schemes outperform the baseline under three linear uplink combining methods and non-linear successive interference cancellation (SIC).

Considering a multi-user Downlink scenario, this paper studies transmit antenna selection (AS) for uniform rectangular arrays (URA). A key contribution of this work is the investigation of the trade-off between AS and the array directivity (or gain), and its impact on spectral efficiency (SE) and energy efficiency (EE). In this context, to improve EE of the system, a directivity-aware transmit AS and power allocation problem is formulated and solved using a sequential approach. For AS, two schemes are proposed: Block selection (BS) and L´evy flight based binary particle swarm optimization (LFBPSO), while for power allocation, an iterative quadratic transform method is employed in combination with Dinkelbach algorithm for EE maximization. The performance of the AS schemes is evaluated with respect to SE, EE and computational complexity. The results show that the array directivity varies with the subset of selected antennas and has a direct impact on the performance of the AS schemes in terms of achievable SE and EE. Furthermore, AS without considering the directivity can incur a loss in the array gain of up to 2 dB in a λ/2 spaced array. The results also show that LFBPSO outperforms other AS schemes with respect to SE and EE. Finally, another important contribution of this work is the experimental validation of the impact of array directivity on the AS process. This is demonstrated by replacing the isotropic antenna array with a measured 4×8 uniform rectangular array of patch antennas in the simulation framework.

This paper introduces a comprehensive framework for assessing the performance of wireless communication systems under mobility conditions. Receiver mobility is modeled using the Random Waypoint (RWP) mobility model. The proposed analysis is applicable to most of the known fading models, provided that the moment generating function (MGF) of the underlying channel fading (static) distribution is readily available. Key performance metrics — such as outage probability, average bit error probability, and ergodic capacity — are represented through a unified integral expression that can be efficiently computed numerically. This framework is versatile and applicable to various practical scenarios, including diversity receivers with arbitrary antenna correlation, wireless communication involving uncrewed aerial vehicles (UAVs), and systems enhanced by intelligent reflecting surfaces (IRS). Numerical results are presented along with Monte Carlo simulations to substantiate the proposed analysis.

In this paper, we focus on the energy efficiency (EE) optimization and analysis of reconfigurable intelligent surface (RIS)-assisted multiuser downlink near-field communications. Specifically, we conduct a comprehensive study on several key factors affecting EE performance, including the number of RIS elements, the types of reconfigurable elements, reconfiguration resolutions, and the maximum transmit power. To accurately capture the power characteristics of RISs, we adopt more practical power consumption models for three commonly used reconfigurable elements in RISs: PIN diodes, varactor diodes, and radio frequency (RF) switches. These different elements may result in RIS systems exhibiting significantly different energy efficiencies (EEs), even when their spectral efficiencies (SEs) are similar. Considering discrete phases implemented at most RISs in practice, which makes their optimization NP-hard, we develop a nested alternating optimization framework to maximize EE, consisting of an outer integer-based optimization for discrete RIS phase reconfigurations and a nested non-convex optimization for continuous transmit power allocation within each iteration. Extensive comparisons with multiple benchmark schemes validate the effectiveness and efficiency of the proposed framework. Furthermore, based on the proposed optimization method, we analyze the EE performance of RISs across different key factors and identify the optimal RIS architecture yielding the highest EE.

In this paper, we focus on the energy efficiency (EE) optimization and analysis of reconfigurable intelligent surface (RIS)-assisted multiuser downlink near-field communications. Specifically, we conduct a comprehensive study on several key factors affecting EE performance, including the number of RIS elements, the types of reconfigurable elements, reconfiguration resolutions, and the maximum transmit power. To accurately capture the power characteristics of RISs, we adopt more practical power consumption models for three commonly used reconfigurable elements in RISs: PIN diodes, varactor diodes, and radio frequency (RF) switches. These different elements may result in RIS systems exhibiting significantly different energy efficiencies (EEs), even when their spectral efficiencies (SEs) are similar. Considering discrete phases implemented at most RISs in practice, which makes their optimization NP-hard, we develop a nested alternating optimization framework to maximize EE, consisting of an outer integer-based optimization for discrete RIS phase reconfigurations and a nested non-convex optimization for continuous transmit power allocation within each iteration. Extensive comparisons with multiple benchmark schemes validate the effectiveness and efficiency of the proposed framework. Furthermore, based on the proposed optimization method, we analyze the EE performance of RISs across different key factors and identify the optimal RIS architecture yielding the highest EE.

Optimizing discrete phase shifts in large-scale reconfigurable intelligent surfaces (RISs) is challenging due to their non-convex and non-linear nature. In this letter, we propose a heuristic-integrated deep reinforcement learning (DRL) framework that (1) leverages accumulated actions over multiple steps in the double deep Q-network (DDQN) for RIS column-wise control and (2) integrates a greedy algorithm (GA) into each DRL step to refine the state via fine-grained, element-wise optimization of RIS configurations. By learning from GA-included states, the proposed approach effectively addresses RIS optimization within a small DRL action space, demonstrating its capability to optimize phase-shift configurations of large-scale RISs.

Optimizing discrete phase shifts in large-scale reconfigurable intelligent surfaces (RISs) is challenging due to their non-convex and non-linear nature. In this letter, we propose a heuristic-integrated deep reinforcement learning (DRL) framework that (1) leverages accumulated actions over multiple steps in the double deep Q-network (DDQN) for RIS column-wise control and (2) integrates a greedy algorithm (GA) into each DRL step to refine the state via fine-grained, element-wise optimization of RIS configurations. By learning from GA-included states, the proposed approach effectively addresses RIS optimization within a small DRL action space, demonstrating its capability to optimize phase-shift configurations of large-scale RISs.

In this study, we provide a theoretical examination of an innovative inverse Class-E power amplifier (PA) incorporating a shunt filter to achieve optimal performance. We derive idealized waveforms and load parameters, verifying these through frequency domain simulations. The findings indicate that the performance of inverse Class-E PA does not consistently increase with higher parasitic output capacitance. The ideal switch voltage and current waveforms suggest the potential for reaching 100% efficiency. We offer design examples using both lumped elements and transmission lines. To validate our theoretical predictions, we conducted measurements on the transmission line inverse Class-E. The simulation and measurement results closely align, showing a maximum drain efficiency of 82%/81% and a power-added efficiency of 75%/74.8%. Furthermore, an output power of 40.5/40.3 dBm and a power gain of 11.0/10.3 dB were achieved in the saturation region.