This paper proposes and evaluates machine learning (ML)-based compensation methods for the transmit (Tx) weight matrices of actual singular value decomposition (SVD)-multiple-input and multiple-output (MIMO) transmissions. These methods train ML models and compensate the Tx weight matrices by using a large amount of training data created from statistical distributions. Moreover, this paper proposes simplified channel metrics based on the channel quality of actual SVD-MIMO transmissions to evaluate compensation performance. The optimal parameters are determined from many ML parameters by using the metrics, and the metrics for this determination are evaluated. Finally, a comprehensive computer simulation shows that the optimal parameters improve performance by up to 7.0dB compared with the conventional method.

Frame synchronization detection before data transmission is an important module which directly affects the lifetime and coexistence of underwater acoustic communication (UAC) networks, where linear frequency modulation (LFM) is a frame preamble signal commonly used for synchronization. Unlike terrestrial wireless communications, strong bursty noise frequently appears in UAC. Due to the long transmission distance and the low signal-to-noise ratio, strong short-distance bursty noise will greatly reduce the accuracy of conventional fractional fourier transform (FrFT) detection. We propose a multi-segment verification fractional fourier transform (MFrFT) preamble detection algorithm to address this challenge. In the proposed algorithm, 4 times of adjacent FrFT operations are carried out. And the LFM signal identifies by observing the linear correlation between two lines connected in pair among three adjacent peak points, called ‘dual-line-correlation mechanism’. The accurate starting time of the LFM signal can be found according to the peak frequency of the adjacent FrFT. More importantly, MFrFT do not result in an increase in computational complexity. Compared with the conventional FrFT detection method, experimental results show that the proposed algorithm can effectively distinguish between signal starting points and bursty noise with much lower error detection rate, which in turn minimizes the cost of retransmission.

This letter presents a dual-band load-modulated sequential amplifier (LMSA). The proposed amplifier changed the attenuator terminated at the isolation port of the four-port combiner of the traditional sequential power amplifier (SPA) architecture into a reactance modulation network (RMN) for load modulation. The impedance can be maintained pure resistance by designing RMN, thus realizing high efficiency and a good portion of the output power in the multiple bands. Compared to the dual-band Doherty power amplifier with a complex dual-band load modulation network (LMN), the proposed LMSA has advantages as maintaining high output power back-off (OBO) efficiency, wide bandwidth and simple construction. A 10-watt dual-band LMSA is simulated and measured in 1.7-1.9GHz and 2.4-2.6GHz with saturated efficiencies 61.2-69.9% and 54.4-70.8%, respectively. The corresponding 9dB OBO efficiency is 46.5-57.1% and 46.4-54.4%, respectively.

One of the performance bottlenecks of a processor is the front-end that supplies instructions. Various techniques, such as cache replacement algorithms and hardware prefetching, have been investigated to facilitate smooth instruction supply at the front-end and to improve processor performance. In these approaches, one of the most important factors has been the reduction in the number of instruction cache misses. By using the number of instruction cache misses or derived factors, previous studies have explained the performance improvements achieved by their proposed methods. However, we found that the number of instruction cache misses does not always explain performance changes well in modern processors. This is because the front-end in modern processors handles subsequent instruction cache misses in overlap with earlier ones. Based on this observation, we propose a novel factor: the number of miss regions. We define a region as a sequence of instructions from one branch misprediction to the next, while we define a miss region as a region that contains one or more instruction cache misses. At the boundary of each region, the pipeline is flushed owing to a branch misprediction. Thus, cache misses after this boundary are not handled in overlap with cache misses before the boundary. As a result, the number of miss regions is equal to the number of cache misses that are processed without overlap. In this paper, we demonstrate that the number of miss regions can well explain the variation in performance through mathematical models and simulation results. The results show that the model explains cycles per instruction with an average error of 1.0% and maximum error of 4.1% when applying an existing prefetcher to the instruction cache. The idea of miss regions highlights that instruction cache misses and branch mispredictions interact with each other in processors with a decoupled front-end. We hope that considering this interaction will motivate the development of fast performance estimation methods and new microarchitectural methods.

In this paper, we apply two methods in machine learning, dropout and semi-supervised learning, to a recently proposed method called CSQ-SDL which uses deep neural networks for evaluating shift quality from time-series measurement data. When developing a new Automatic Transmission (AT), calibration takes place where many parameters of the AT are adjusted to realize pleasant driving experience in all situations that occur on all roads around the world. Calibration requires an expert to visually assess the shift quality from the time-series measurement data of the experiments each time the parameters are changed, which is iterative and time-consuming. The CSQ-SDL was developed to shorten time consumed by the visual assessment, and its effectiveness depends on acquiring a sufficient number of data points. In practice, however, data amounts are often insufficient. The methods proposed here can handle such cases. For the cases wherein only a small number of labeled data points is available, we propose a method that uses dropout. For those cases wherein the number of labeled data points is small but the number of unlabeled data is sufficient, we propose a method that uses semi-supervised learning. Experiments show that while the former gives moderate improvement, the latter offers a significant performance improvement.

Recently, a lot of low-light image enhancement methods have been proposed. However, these methods have some problems such as causing fine details lost in bright regions and/or unnatural color tones. In this paper, we propose a new low-light image enhancement method to cope with these problems. In the proposed method, a pixel is represented by a convex combination of white, black, and pure color. Then, an equi-hue plane in RGB color space is represented as a triangle whose vertices correspond to white, black, and pure color. The visibility of low-light image is improved by applying a modified gamma transform to the combination coefficients on an equi-hue plane in RGB color space. The contrast of the image is enhanced by the histogram specification method using the histogram smoothed by a filter with a kernel determined based on a gamma distribution. In the experiments, the effectiveness of the proposed method is verified by the comparison with the state-of-the-art low-light image enhancement methods.

We propose a non-photorealistic rendering method for automatically generating point-light-style (PLS) images from photographic images using peripheral difference filters with different window sizes. The proposed method can express PLS patterns near the edges of photographic images as dots. To verify the effectiveness of the proposed method, experiments were conducted to visually confirm PLS images generated from various photographic images.

Most quantum circuits require SWAP gate insertion to run on quantum hardware with limited qubit connectivity. A promising SWAP gate insertion method for blocks of commuting two-qubit gates is a predetermined swap strategy which applies layers of SWAP gates simultaneously executable on the coupling map. A good initial mapping for the swap strategy reduces the number of required swap gates. However, even when a circuit consists of commuting gates, e.g., as in the Quantum Approximate Optimization Algorithm (QAOA) or trotterized simulations of Ising Hamiltonians, finding a good initial mapping is a hard problem. We present a SAT-based approach to find good initial mappings for circuits with commuting gates transpiled to the hardware with swap strategies. Our method achieves a 65% reduction in gate count for random three-regular graphs with 500 nodes. In addition, we present a heuristic approach that combines the SAT formulation with a clustering algorithm to reduce large problems to a manageable size. This approach reduces the number of swap layers by 25% compared to both a trivial and random initial mapping for a random three-regular graph with 1000 nodes. Good initial mappings will therefore enable the study of quantum algorithms, such as QAOA and Ising Hamiltonian simulation applied to sparse problems, on noisy quantum hardware with several hundreds of qubits.

We thank Kamata et al. (2023) [1] for their interest in our work [2], and for providing an explanation of the quasi-linear kernel from a viewpoint of multiple kernel learning. In this letter, we first give a summary of the quasi-linear SVM. Then we provide a discussion on the novelty of quasi-linear kernels against multiple kernel learning. Finally, we explain the contributions of our work [2].

We compare the demodulation performance of an analog OTDM demultiplexing scheme and digitized OTDM demultiplexing with an ultrahigh-speed digital signal processor in a single-channel OTDM coherent Nyquist pulse transmission. We evaluated the demodulation performance for 40, 80, and 160Gbaud OTDM signals with a baseline rate of 10Gbaud. As a result, we clarified that the analog scheme performs significantly better since the bandwidth for handling the demultiplexed signal is as narrow as 10GHz regardless of the symbol rate. This enables us to use a low-speed A/D converter (ADC) with a large effective number of bits (ENOB). On the other hand, in the digital scheme, the higher the symbol rate becomes, the more bandwidth the receiver requires. Therefore, it is necessary to use an ultrahigh-speed ADC with a low ENOB for a 160Gbaud signal. We measured the ENOB of the ultrahigh-speed ADC used in the digital scheme and showed that the measured ENOB was approximately 1.5 bits lower than that of the low-speed ADC used in the analog scheme. This 1.5-bit decrease causes a large degradation in the demodulation performance obtained with the digital demultiplexing scheme.

The Common Media Application Format (CMAF) is a standard for adaptive bitrate live streaming. The CMAF adapts chunk encoding and enables low-latency live streaming. However, conventional bandwidth estimation for adaptive bitrate streaming underestimates bandwidth because download time is affected not only by network bandwidth but also by the idle times between chunks in the same segment. Inaccurate bandwidth estimation decreases the quality of experience of the streaming client. In this paper, we propose a chunk-grouping method to estimate the available bandwidth for adaptive bitrate live streaming. In the proposed method, by delaying HTTP request transmission and bandwidth estimation using grouped chunks, the client estimates the available bandwidth accurately due to there being no idle times in the grouped chunks. In addition, we extend the proposed method to dynamically change the number of grouping chunks according to buffer length during downloading of the previous segment. We evaluate the proposed methods under various network conditions in order to confirm the effectiveness of the proposed methods.

In this study, an overloaded large-scale distributed antenna network is considered, for which the number of active users is larger than that of antennas distributed in a base station coverage area (called a cell). To avoid overload, users in each cell are divided into multiple user groups, and, to reduce the computational complexity required for multi-user multiple-input and multiple-output (MU-MIMO), users in each user group are grouped into multiple user clusters so that cluster-wise distributed MU-MIMO can be performed in parallel in each user group. However, as the network size increases, conventional computational methods may not be able to solve combinatorial optimization problems, such as user scheduling and user clustering, which are required for performing cluster-wise distributed MU-MIMO in a finite amount of time. In this study, we apply quantum computing to solve the combinatorial optimization problems of user scheduling and clustering for an overloaded distributed antenna network and propose a quantum computing-based user scheduling and clustering method. The results of computer simulations indicate that as the technology of quantum computers and their related algorithms evolves in the future, the proposed method can realize large-scale dense wireless systems and realize real-time optimization with a short optimization execution cycle.

This paper extends our previously reported non-orthogonal multiple access (NOMA)-based highly-efficient and low-latency hybrid automatic repeat request (HARQ) method for ultra-reliable low latency communications (URLLC) to the case with inter-base station cooperation. In the proposed method, delay-sensitive URLLC packets are preferentially multiplexed with best-effort enhanced mobile broadband (eMBB) packets in the same channel using superposition coding to reduce the transmission latency of the URLLC packet while alleviating the throughput loss in eMBB. Although data transmission to the URLLC terminal is conducted by multiple base stations based on inter-base station cooperation, the proposed method allocates radio resources to URLLC terminals which include scheduling (bandwidth allocation) and power allocation at each base station independently to achieve the short transmission latency required for URLLC. To avoid excessive radio resource assignment to URLLC terminals due to independent resource assignment at each base station, which may result in throughput degradation in eMBB terminals, we employ an adaptive path-loss-dependent weighting approach in the scheduling-metric calculation. This achieves appropriate radio resource assignment to URLLC terminals while reducing the packet error rate (PER) and transmission delay time thanks to the inter-base station cooperation. We show that the proposed method significantly improves the overall performance of the system that provides simultaneous eMBB and URLLC services.

In this study, we propose a method for localizing an unknown moving emitter by measuring a sequence of the frequency-of-arrival using a single moving observation platform. Furthermore, we introduce the position and velocity errors of the moving observation platform into the theoretical localization error equation to analyze the effect of these errors on the localization accuracy without Monte-Carlo simulations. The proposed theoretical error equation can propagate toward the time direction; therefore, the theoretical localization error can be evaluated at an arbitral time. We demonstrate that the localization error value obtained by the proposed equation and the RMSE evaluated by the Monte-Carlo simulation sufficiently coincide with one another.

Recently, “Both-Side Retrodirective System” was proposed, as a beam convergence technique, for microwave high power transmission. To demonstrate the effectiveness of the both-side retrodirective system by experiment, the authors propose a 2-dimensional measurement equipment. Propagation in the parallel plate waveguide was analogized based on free-space propagation, and the theory and characteristics were clarified by simulation. The electric field distribution in the waveguide was measured by electric probe with the proposed equipment. Two types of measurement equipment were developed. One is a 4-element experiment system, which is a small-scale device for principle verification. The other is a 16-element measurement equipment, which is intended to evaluate beam convergence of a both-side retrodirective system in the next step. The measured results were compared with simulation results. As a result, it was confirmed that the beam formed in the waveguide was successfully measured. Thus, the effectiveness of 2-dimensional measurement equipment for evaluation of beam convergence was shown.

We propose a novel silicon diplexer integrated with filters for frequency-division multiplexing in the 300-GHz band. The diplexer consists of a directional coupler formed of unclad silicon wires, a photonic bandgap-based low-pass filter, and a high-pass filter based on frequency-dependent bending loss. These integrated filters are capable of suppressing crosstalk and providing >15dB isolation over 40GHz, which is highly beneficial for terahertz-range wireless communications applications. We have used this diplexer in a simultaneous error-free wireless transmission of 300-GHz and 335-GHz channels at the aggregate data rate of 36Gbit/s.

The Variational Quantum Eigensolver (VQE) algorithm is gaining interest for its potential use in near-term quantum devices. In the VQE algorithm, parameterized quantum circuits (PQCs) are employed to prepare quantum states, which are then utilized to compute the expectation value of a given Hamiltonian. Designing efficient PQCs is crucial for improving convergence speed. In this study, we introduce problem-specific PQCs tailored for optimization problems by dynamically generating PQCs that incorporate problem constraints. This approach reduces a search space by focusing on unitary transformations that benefit the VQE algorithm, and accelerate convergence. Our experimental results demonstrate that the convergence speed of our proposed PQCs outperforms state-of-the-art PQCs, highlighting the potential of problem-specific PQCs in optimization problems.

FPGA clusters that consist of multiple FPGA boards have been gaining interest in recent times. Massively parallel processing with a stand-alone heterogeneous FPGA cluster with SoC- style FPGAs and mid-scale FPGAs is promising with cost-performance benefit. Here, we propose such a heterogeneous FPGA cluster with FiC and M-KUBOS cluster. FiC consists of multiple boards, mounting middle scale Xilinx's FPGAs and DRAMs, which are tightly coupled with high-speed serial links. In addition, M-KUBOS boards are connected to FiC for ensuring high IO data transfer bandwidth. As an example of massively parallel processing, here we implement genomic pattern search. Next-generation sequencing (NGS) technology has revolutionized biological system related research by its high-speed, scalable and massive throughput. To analyze the genomic data, short read mapping technique is used where short Deoxyribonucleic acid (DNA) sequences are mapped relative to a known reference sequence. Although several pattern matching techniques are available, FM-index based pattern search is perfectly suitable for this task due to the fastest mapping from known indices. Since matching can be done in parallel for different data, the massively parallel computing which distributes data, executes in parallel and gathers the results can be applied. We also implement a data compression method where about 10 times reduction in data size is achieved. We found that a M-KUBOS board matches four FiC boards, and a system with six M-KUBOS boards and 24 FiC boards achieved 30 times faster than the software based implementation.

This letter presents a global feature-based method for evaluating the no reference quality of scanning electron microscopy (SEM) contrast-distorted images. Based on the characteristics of SEM images and the human visual system, the global features of SEM images are extracted as the score for evaluating image quality. In this letter, the texture information of SEM images is first extracted using a low-pass filter with orientation, and the amount of information in the texture part is calculated based on the entropy reflecting the complexity of the texture. The singular values with four scales of the original image are then calculated, and the amount of structural change between different scales is calculated and averaged. Finally, the amounts of texture information and structural change are pooled to generate the final quality score of the SEM image. Experimental results show that the method can effectively evaluate the quality of SEM contrast-distorted images.

Orthogonal frequency division multiplexing (OFDM) is very sensitive to the carrier frequency offset (CFO). The CFO estimation precision heavily makes impacts on the OFDM performance. In this paper, a new Bayesian learning-assisted joint CFO tracking and channel impulse response estimation is proposed. The proposed algorithm is modified from a Bayesian learning-assisted estimation (BLAE) algorithm in the literature. The BLAE is expectation-maximization (EM)-based and displays the estimator mean square error (MSE) lower than the Cramer-Rao bound (CRB) when the CFO value is near zero. However, its MSE value may increase quickly as the CFO value goes away from zero. Hence, the CFO estimator of the BLAE is replaced to solve the problem. Originally, the design criterion of the single-time-sample (STS) CFO estimator in the literature is maximum likelihood (ML)-based. Its MSE performance can reach the CRB. Also, its CFO estimation range can reach the widest range required for a CFO tracking estimator. For a CFO normalized by the sub-carrier spacing, the widest tracking range required is from -0.5 to +0.5. Here, we apply the STS CFO estimator design method to the EM-based Bayesian learning framework. The resultant Bayesian learning-assisted STS algorithm displays the MSE performance lower than the CRB, and its CFO estimation range is between ±0.5. With such a Bayesian learning design criterion, the additional channel noise power and power delay profile must be estimated, as compared with the ML-based design criterion. With the additional channel statistical information, the derived algorithm presents the MSE performance better than the CRB. Two frequency-selective channels are adopted for computer simulations. One has fixed tap weights, and the other is Rayleigh fading. Comparisons with the most related algorithms are also been provided.