Masahiro YOKOTA Kazuhisa YAMAGISHI
In this paper, the quality and transferred data based video bitrate control method for web-conferencing services is proposed, aiming to reduce transferred data by suppressing excessive quality. In web-conferencing services, the video bitrate is generally controlled in accordance with the network conditions (e.g., jitter and packet loss rate) to improve users' quality. However, in such a control, the bitrate is excessively high when the network conditions is sufficiently high (e.g., high throughput and low jitter), which causes an increased transferred data volume. The increased volume of data transferred leads to increased operational costs, such as network costs for service providers. To solve this problem, we developed a method to control the video bitrate of each user to achieve the required quality determined by the service provider. This method is implemented in an actual web-conferencing system and evaluated under various conditions. It was shown that the bitrate could be controlled in accordance with the required quality to reduce the transferred data volume.
Fujihiko MATSUMOTO Hinano OHTSU
In a field of biomedical engineering, not only low-pass filters for high frequency elimination but also notch filters for suppressing powerline interference are necessary to process low-frequency biosignals. For integration of low-frequency filters, chip implementation of large capacitances is major difficulty. As methods to enhance capacitances with small chip area, use of capacitance multipliers is effective. This letter describes design consideration of integrated low-frequency low-pass notch filter employing capacitance multipliers. Two main points are presented. Firstly, a new floating capacitance multiplier is proposed. Secondly, a technique to reduce the number of capacitance multipliers is proposed. By this technique, power consumption is reduced. The proposed techniques are applied a 3rd order low-pass notch filter. Simulation results show the effectiveness of the proposed techniques.
Aorui GOU Jingjing LIU Xiaoxiang CHEN Xiaoyang ZENG Yibo FAN
Convolutional Neural Networks (CNNs) and Transformers have achieved remarkable performance in detection and classification tasks. Nevertheless, their feature extraction cannot consider both local and global information, so the detection and classification performance can be further improved. In addition, more and more deep learning networks are designed as more and more complex, and the amount of computation and storage space required is also significantly increased. This paper proposes a combination of CNN and transformer, and designs a local feature enhancement module and global context modeling module to enhance the cascade network. While the local feature enhancement module increases the range of feature extraction, the global context modeling is used to capture the feature maps' global information. To decrease the model complexity, a shared sublayer is designed to realize the sharing of weight parameters between the adjacent convolutional layers or cross convolutional layers, thereby reducing the number of convolutional weight parameters. Moreover, to effectively improve the detection performance of neural networks without increasing network parameters, the optimal transport assignment approach is proposed to resolve the problem of label assignment. The classification loss and regression loss are the summations of the cost between the demander and supplier. The experiment results demonstrate that the proposed Combination of CNN and Transformer with Shared Sublayer (CCTSS) performs better than the state-of-the-art methods in various datasets and applications.
Chaorong ZHANG Yuyang PENG Ming YUE Fawaz AL-HAZEMI
As a potential member of next generation wireless communications, the reconfigurable intelligent surface (RIS) can control the reflected elements to adjust the phase of the transmitted signal with less energy consumption. A novel RIS-assisted index modulation scheme is proposed in this paper, which is named the generalized reflected phase modulation (GRPM). In the GRPM, the transmitted bits are mapped into the reflected phase combination which is conveyed through the reflected elements on the RIS, and detected by the maximum likelihood (ML) detector. The performance analysis of the GRPM with the ML detector is presented, in which the closed form expression of pairwise error probability is derived. The simulation results show the bit error rate (BER) performance of GRPM by comparing with various RIS-assisted index modulation schemes in the conditions of various spectral efficiency and number of antennas.
Kenji KANAI Hidehiro KANEMITSU Taku YAMAZAKI Shintaro MORI Aram MINE Sumiko MIYATA Hironobu IMAMURA Hidenori NAKAZATO
A city-level digital twin is a critical enabling technology to construct a smart city that helps improve citizens' living conditions and quality of life. Currently, research and development regarding the digital replica city are pursued worldwide. However, many research projects only focus on creating the 3D city model. A mechanism to involve key players, such as data providers, service providers, and application developers, is essential for constructing the digital replica city and producing various city applications. Based on this motivation, the authors of this paper are pursuing a research project, namely Decentralized Digital Twin EcoSystem (D2EcoSys), to create an ecosystem to advance (and self-grow) the digital replica city regarding time and space directions, city services, and values. This paper introduces an overview of the D2EcoSys project: vision, problem statement, and approach. In addition, the paper discusses the recent research results regarding networking technologies and demonstrates an early testbed built in the Kashiwa-no-ha smart city.
Hiroki URASAWA Hayato SOYA Kazuhiro YAMAGUCHI Hideaki MATSUE
We evaluated the transmission performance, including received power and transmission throughput characteristics, in 4×4 single-user multiple-input multiple-output (SU-MIMO) transmission for synchronous time division duplex (TDD) and downlink data channels in comparison with single-input single-output (SISO) transmission in an environment where a local 5G wireless base station was installed on the roof of a research building at our university. Accordingly, for the received power characteristics, the difference between the simulation value, which was based on the ray tracing method, and the experimental value at 32 points in the area was within a maximum difference of approximately 10 dB, and sufficient compliance was obtained. Regarding the transmission throughput versus received power characteristics, after showing a simulation method for evaluating throughput characteristics in MIMO, we compared the results with experimental results. The cumulative distribution function (CDF) of the transmission throughput shows that, at a CDF of 50%, in SISO transmission, the simulated value is approximately 115Mbps, and the experimental value is 105Mbps, within a difference of approximately 10Mbps. By contrast, in MIMO transmission, the simulation value is 380Mbps, and the experimental value is approximately 420Mbps, which is a difference of approximately 40Mbps. It was shown that the received power and transmission throughput characteristics can be predicted with sufficient accuracy by obtaining the delay profile and the system model at each reception point using the both ray tracing and MIMO simulation methods in actual environments.
Ting WANG Xianhua NIU Yaoxuan WANG Jianhong ZHOU Ling XIONG
The frequency hopping sequence plays a crucial role in determining the system's anti-jamming performance, in frequency hopping communication systems. If the adjacent frequency points of FHS can ensure wide-gap, it will better improve the anti-interference capability of the FH communication system. Moreover, if the period of the sequence is expanded, and each frequency point does not repeat in the same sequence, the system's ability to resist electromagnetic interference will be enhanced. And a one-coincidence frequency-hopping sequence set consists of FHSs with maximum Hamming autocorrelation 0 and cross-correlation 1. In this paper, we present two constructions of wide-gap frequency-hopping sequence sets. One construction is a new class of wide-gap one-coincidence FHS set, and the other is a WGFHS set with long period. These two WGFHS sets are optimal with respect to WG-Peng-Fan bound. And each sequence of these WGFHS sets is optimal with respect to WG-Lempel-Greenberger bound.
Chenchen LIU Wenyi ZHANG Xiaoni DU
The calculation of cross-correlation between a sequence with good autocorrelation and its decimated sequence is an interesting problem in the field of sequence design. In this letter, we consider a class of ternary sequences with perfect autocorrelation, proposed by Shedd and Sarwate (IEEE Trans. Inf. Theory, 1979, DOI: 10.1109/TIT.1979.1055998), which is generated based on the cross-correlation between m-sequence and its d-decimation sequence. We calculate the cross-correlation distribution between a certain pair of such ternary perfect sequences and show that the cross-correlation takes three different values.
Yu ZHOU Jianyong HU Xudong MIAO Xiaoni DU
Low confusion coefficient values can make side-channel attacks harder for vector Boolean functions in Block cipher. In this paper, we give new results of confusion coefficient for f ⊞ g, f ⊡ g, f ⊕ g and fg for different Boolean functions f and g, respectively. And we deduce a relationship on the sum-of-squares of the confusion coefficient between one n-variable function and two (n - 1)-variable decomposition functions. Finally, we find that the confusion coefficient of vector Boolean functions is affine invariant.
A group signature scheme allows us to anonymously sign a message on behalf of a group. One of important issues in the group signatures is user revocation, and thus lots of revocable group signature (RGS) schemes have been proposed so far. One of the applications suitable to the group signature is privacy-enhancing crowdsensing, where the group signature allows mobile sensing users to be anonymously authenticated to hide the location. In the mobile environment, verifier-local revocation (VLR) type of RGS schemes are suitable, since revocation list (RL) is not needed in the user side. However, in the conventional VLR-RGS schemes, the revocation check in the verifier needs O(R) cryptographic operations for the number R of revoked users. On this background, VLR-RGS schemes with efficient revocation check have been recently proposed, where the revocation check is just (bit-string) matching. However, in the existing schemes, signatures are linkable in the same interval or in the same application-independent task with a public index. The linkability is useful in some scenarios, but users want the unlinkability for the stronger anonymity. In this paper, by introducing a property that at most K unlinkable signatures can be issued by a signer during each interval for a fixed integer K, we propose a VLR-RGS scheme with the revocation token matching. In our scheme, even the signatures during the same interval are unlinkable. Furthermore, since used indexes are hidden, the strong anonymity remains. The overheads are the computational costs of the revocation algorithm and the RL size. We show that the overheads are practical in use cases of crowdsensing.
Atsushi TAGAMI Takuya MIYASAKA Masaki SUZUKI Chikara SASAKI
Recently, there has been a surge of interest in Artificial Intelligence (AI) and its applications have been considered in various fields. Mobile networks are becoming an indispensable part of our society, and are considered as one of the promising applications of AI. In the Beyond 5G/6G era, AI will continue to penetrate networks and AI will become an integral part of mobile networks. This paper provides an overview of the collaborations between networks and AI from two categories, “AI for Network” and “Network for AI,” and predicts mobile networks in the B5G/6G era. It is expected that the future mobile network will be an integrated infrastructure, which will not only be a mere application of AI, but also provide as the process infrastructure for AI applications. This integration requires a driving application, and the network operation is one of the leading candidates. Furthermore, the paper describes the latest research and standardization trends in the autonomous networks, which aims to fully automate network operation, as a future network operation concept with AI, and discusses research issues in the future mobile networks.
Satoshi YONEDA Akihito KOBAYASHI Eiji TANIGUCHI
An ESL-cancelling circuit for a shunt-connected film capacitor filter using vertically stacked coupled square loops is reported in this paper. The circuit is applicable for a shunt-connected capacitor filter whose equivalent series inductance (ESL) of the shunt-path causes deterioration of filter performance at frequencies above the self-resonant frequency. Two pairs of vertically stacked magnetically coupled square loops are used in the circuit those can equivalently add negative inductance in series to the shunt-path to cancel ESL for improvement of the filter performance. The ESL-cancelling circuit for a 1-μF film capacitor was designed according to the Biot-Savart law and electromagnetic (EM)-analysis, and the prototype was fabricated with an FR4 substrate. The measured result showed 20-dB improvement of the filter performance above the self-resonant frequency as designed, satisfying Sdd21 less than -40dB at 1MHz to 100MHz. This result is almost equivalent to reduce ESL of the shunt-path to less than 1nH at 100MHz and is also difficult to realize using any kind of a single bulky film capacitor without cancelling ESL.
Rin OISHI Junichiro KADOMOTO Hidetsugu IRIE Shuichi SAKAI
As more and more programs handle personal information, the demand for secure handling of data is increasing. The protocol that satisfies this demand is called Secure function evaluation (SFE) and has attracted much attention from a privacy protection perspective. In two-party SFE, two mutually untrustworthy parties compute an arbitrary function on their respective secret inputs without disclosing any information other than the output of the function. For example, it is possible to execute a program while protecting private information, such as genomic information. The garbled circuit (GC) — a method of program obfuscation in which the program is divided into gates and the output is calculated using a symmetric key cipher for each gate — is an efficient method for this purpose. However, GC is computationally expensive and has a significant overhead even with an accelerator. We focus on hardware acceleration because of the nature of GC, which is limited to certain types of calculations, such as encryption and XOR. In this paper, we propose an architecture that accelerates garbling by running multiple garbling engines simultaneously based on the latest FPGA-based GC accelerator. In this architecture, managers are introduced to perform multiple rows of pipeline processing simultaneously. We also propose an optimized implementation of RAM for this FPGA accelerator. As a result, it achieves an average performance improvement of 26% in garbling the same set of programs, compared to the state-of-the-art (SOTA) garbling accelerator.
Siyi HU Makiko ITO Takahide YOSHIKAWA Yuan HE Hiroshi NAKAMURA Masaaki KONDO
Widely adopted by machine learning and graph processing applications nowadays, sparse matrix-Vector multiplication (SpMV) is a very popular algorithm in linear algebra. This is especially the case for fully-connected MLP layers, which dominate many SpMV computations and play a substantial role in diverse services. As a consequence, a large fraction of data center cycles is spent on SpMV kernels. Meanwhile, despite having efficient storage options against sparsity (such as CSR or CSC), SpMV kernels still suffer from the problem of limited memory bandwidth during data transferring because of the memory hierarchy of modern computing systems. In more detail, we find that both integer and floating-point data used in SpMV kernels are handled plainly without any necessary pre-processing. Therefore, we believe bandwidth conservation techniques, such as data compression, may dramatically help SpMV kernels when data is transferred between the main memory and the Last Level Cache (LLC). Furthermore, we also observe that convergence conditions in some typical scientific computation benchmarks (based on SpMV kernels) will not be degraded when adopting lower precision floating-point data. Based on these findings, in this work, we propose a simple yet effective data compression scheme that can be extended to general purpose computing architectures or HPC systems preferably. When it is adopted, a best-case speedup of 1.92x is made. Besides, evaluations with both the CG kernel and the PageRank algorithm indicate that our proposal introduces negligible overhead on both the convergence speed and the accuracy of final results.
Jianbo WANG Haozhi HUANG Li SHEN Xuan WANG Toshihiko YAMASAKI
The image-to-image translation aims to learn a mapping between the source and target domains. For improving visual quality, the majority of previous works adopt multi-stage techniques to refine coarse results in a progressive manner. In this work, we present a novel approach for generating plausible details by only introducing a group of intermediate supervisions without cascading multiple stages. Specifically, we propose a Laplacian Pyramid Transformation Generative Adversarial Network (LapTransGAN) to simultaneously transform components in different frequencies from the source domain to the target domain within only one stage. Hierarchical perceptual and gradient penalization are utilized for learning consistent semantic structures and details at each pyramid level. The proposed model is evaluated based on various metrics, including the similarity in feature maps, reconstruction quality, segmentation accuracy, similarity in details, and qualitative appearances. Our experiments show that LapTransGAN can achieve a much better quantitative performance than both the supervised pix2pix model and the unsupervised CycleGAN model. Comprehensive ablation experiments are conducted to study the contribution of each component.
Mashiho MUKAIDA Yoshiaki UEDA Noriaki SUETAKE
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.
Gouki OKADA Makoto NAKASHIZUKA
This paper presents a deep network based on unrolling the diffusion process with the morphological Laplacian. The diffusion process is an iterative algorithm that can solve the diffusion equation and represents time evolution with Laplacian. The diffusion process is applied to smoothing of images and has been extended with non-linear operators for various image processing tasks. In this study, we introduce the morphological Laplacian to the basic diffusion process and unwrap to deep networks. The morphological filters are non-linear operators with parameters that are referred to as structuring elements. The discrete Laplacian can be approximated with the morphological filters without multiplications. Owing to the non-linearity of the morphological filter with trainable structuring elements, the training uses error back propagation and the network of the morphology can be adapted to specific image processing applications. We introduce two extensions of the morphological Laplacian for deep networks. Since the morphological filters are realized with addition, max, and min, the error caused by the limited bit-length is not amplified. Consequently, the morphological parts of the network are implemented in unsigned 8-bit integer with single instruction multiple data set (SIMD) to achieve fast computation on small devices. We applied the proposed network to image completion and Gaussian denoising. The results and computational time are compared with other denoising algorithm and deep networks.
Atsushi MATSUO Shigeru YAMASHITA Daniel J. EGGER
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.
In this paper, a circuit based on a field programmable analog array (FPAA) is proposed for three types of chaotic spiking oscillator (CSO). The input/output conversion characteristics of a specific element in the FPAA can be defined by the user. By selecting the proper characteristics, three types of CSO are realized without changing the structure of the circuit itself. Chaotic attractors are observed in a hardware experiment. It is confirmed that the dynamics of the CSOs are consistent with numerical simulations.
Yuichiro NISHIKAWA Shota NISHIJIMA Akira HIRANO
We have proposed autonomous network diagnosis platform for operation of future large capacity and virtualized network, including 5G and beyond 5G services. As for the one candidate of information collection and analyzing function blocks in the platform, we proposed novel optical sensing techniques that utilized tapped raw signal data acquired from digital coherent optical receivers. The raw signal data is captured before various digital signal processing for demodulation. Therefore, it contains various waveform deformation and/or noise as it experiences through transmission fibers. In this paper, we examined to detect two possible failures in transmission lines including fiber bending and optical filter shift by analyzing the above-mentioned raw signal data with the help of machine learning. For the purpose, we have implemented Docker container applications in WhiteBox Cassini to acquire real-time raw signal data. We generated CNN model for the detections in off-line processing and used them for real-time detections. We have confirmed successful detection of optical fiber bend and/or optical filter shift in real-time with high accuracy. Also, we evaluated their tolerance against ASE noise and invented novel approach to improve detection accuracy. In addition to that, we succeeded to detect them even in the situation of simultaneous occurrence of those failures.