Published papers

We propose a bandwidth allocation technique for accommodating a time division duplex mobile base station (TDD-BS) and a secondary system in a TDM-PON. Operating availability is confirmed with essential three types of TDD configuration taken from the seven types described by the Third Generation Partnership Project (3GPP) standardization group. We evaluate the latency of the TDD-BS and the throughput of the secondary system. We report that the TDD-BS signal is forwarded in less than about 50 µs and is unaffected by the secondary system. The throughput of the secondary system is improved compared with that of the conventional technique.

This paper employs representative machine learning techniques on on-off encoded eigenvalue demodulation and investigates the optimum in terms of BER performance. Simulation results indicated a random forest technique is the best in the 10-eigenvalue case.

This paper employs representative machine learning techniques on on-off encoded eigenvalue demodulation and investigates the optimum in terms of BER performance. Simulation results indicated a random forest technique is the best in the 10-eigenvalue case.

This paper investigates transmission characteristics in radio over fiber systems with linear array antennas in terms of the interaction between fiber nonlinearity and chromatic dispersion. We report that O-band has better EVM characteristics than C-band.

This paper investigates transmission characteristics in radio over fiber systems with linear array antennas in terms of the interaction between fiber nonlinearity and chromatic dispersion. We report that O-band has better EVM characteristics than C-band.

Introduction: Optical eigenvalue communication systems use eigenvalues extracted from a matrix calculated by inverse scattering transform (IST) [1]. For on-off encoded eigenvalue modulation schemes, which allocate the on-off state of each eigenvalue to each bit, it is helpful to employ the neural network (NN) method on demodulation [2]. To reduce the time complexity to extract eigenvalues, we have proposed a demodulation scheme employing NN and a complex moment eigenvalue (CME) solver [3]. CME consists of a two-stage algorithm; the first is an estimation method of the number of eigenvalues in the preset arbitrary regions, and the second is the eigenvalue extraction based on the first-stage result. To further reduce complexity, this paper proposes a demodulation scheme using the NN and the estimation the number of eigenvalues, the 1st stage of CME, and the experiments report bit error rate (BER) characteristics compared with the related methods.

This paper proposes a novel return light cancellation technique using an adaptive interference equalizer in the wavelength division multiplexed passive optical network (WDM-PON) transmission system employing the single-wavelength bidirectional transmission scheme. In this scheme, the same wavelength is assigned to both the uplink and the downlink. In other words, the number of the wavelengths allocated to optical network units can be half compared with a typical PON system. However, this presents us with a problem in that the reflected component of the transmitted signal light interferes with the received signal light. We investigated the performance of the proposed interference canceling equalizer using back-to-back transmission simulation and a dual-polarization-quadrature-phase-shift-keying signal transmission experiment. The results indicated that the proposed adaptive equalizer enabled reception with no error when the OSNR was sufficiently large, even when the power of the interfering light was large (i.e., SIR geq 2 dB).

This paper proposes optimizing the compression rate of segmented image region in deep joint source-channel coding (DeepJSCC) transmission in order to improve both image quality and communication efficiency. When transmitting highresolution images with DeepJSCC, segmented transmission is employed due to computational constraints. Additionally, the structure of the output layer of the convolutional neural network (CNN) can adjust its number of transmitted symbols for each region. Focusing on this feature, we attempt to optimize the compression rate for each region of the image based on the segment's entropy value. Its validity is examined through singleinput single-output (SISO) and multiple-input multiple-output (MIMO) channels demonstrating improvements in decoding accuracy of target regions while maintaining overall image quality, especially in low-SNR environments. Beyond the entropy-based importance classification used in this paper, the potential for further accuracy improvements through more precise importance classification is also suggested.

Deep Joint Source-Channel Coding (DeepJSCC) has gained attention as a form of semantic communication that conveys not only information but also meaning and intent. It leverages deep learning, utilizing an autoencoder to map information sources such as images directly to IQ symbols. DeepJSCC has been extensively studied for image transmission, demonstrating advantages such as avoiding the cliff effect and achieving a high Peak Signal-to-Noise Ratio (PSNR) even in low Signal-to-Noise Ratio (SNR) regions. Traditionally, convolutional neural network (CNN)-based autoencoders are used in DeepJSCC for image transmission. However, depending on the terminal used, these methods can be computationally intensive. To address this challenge, the authors have proposed FFT-DeepJSCC, which replaces the 2D convolutional layer within the CNN with Fast Fourier Transform (FFT) and element-wise product operations. This modification aims to reduce the computational burden while maintaining performance. The study evaluates the computational load reduction achieved by FFT-DeepJSCC and examines the impact of the altered layer structure on PSNR. Furthermore, the authors validate the method's effectiveness through actual image transmission experiments using a modified commercially available 5G base station and user terminal. The results demonstrate the performance and feasibility of the proposed approach.

This paper proposes a simplified enhancement of deep learning-based Joint Source-Channel Coding (Deep JSCC) over multiple-input multiple-output (MIMO) channel. Deep JSCC utilizes trained encoder and decoder weights, allowing for the integrated implementation of source coding and channel coding for batch image transmission. During the training phase, the system is trained based on single-input single-output (SISO) with additive white Gaussian noise (AWGN) communication channel principles. In the testing phase, image signals are transmitted via MIMO channels after passing through the respective Deep JSCC encoder and precoding, and subsequently, these signals are input to the decoder after MIMO detection. We examine fundamental image transmission performance in terms of PSNR and SSIM under various MIMO weight designs.

The effect of fiber nonlinearity can be an impairment in terahertz analog radio-over-fiber (A-RoF) systems that require higher optical power and numerous channel multiplexing operations. In this study, we propose a novel precompensation method for nonlinear distortion in A-RoF transmissions and demonstrate the validity of the proposed method through numerical simulations.

Underwater communication is one of the biggest challenges for 6G communications to provide ubiquitous connectivity all over the world. Although optical communication is a strong option for underwater communication, the directivity control of laser beams has been a considerable practical issue due to oceanic turbulence. While optical camera communication (OCC) is an emerging technology for next generation wireless communication, there have been few works on underwater OCC (UWOCC). The propagation characteristics of the optical signals transmitted from LED lights in UWOCC have not been well studied. Therefore, This paper proposes a predictive equalization technique for UWOCC assuming color shift keying (CSK), where the optical signals are modulated by modifying the intensity of the three-color LED luminaires. The proposed technique predictively equalizes the received signals from the symbol color and link distance leveraging the different attenuation of light intensity depending on the wavelength. We demonstrate the feasibility of the proposed idea via experimental results at a depth of 3.5 meters. The proposed equalization improved symbol error rate (SER) so that the overall bit error rate (BER) was significantly suppressed.

It has been a hot research topic to detect and mitigate Distributed Denial-of-Service (DDoS) attacks due to the significant increase of serious threat of such attacks. The rapid growth of Internet of Things (IoT) has intensified this trend, e.g. the Mirai botnet and variants. To address this issue, a light-weight DDoS mitigation mechanism was presented. In the proposed scheme, flooding attacks are detected by stochastic queue allocation which can be executed with widespread and inexpensive commercial products at a network edge. However, the detection process is delayed when the number of incoming flows is large because of the randomness of queue allocation. Thus, in this paper we propose an efficient queue allocation algorithm for rapid DDoS mitigation using limited resources. The idea behind the proposed scheme is to avoid duplicate allocation by decreasing the randomness of the existing scheme. The performance of the proposed scheme was confirmed via theoretical analysis and computer simulation. As a result, it was confirmed that malicious flows are efficiently detected and discarded with the proposed algorithm.