Published papers

This study proposes combining the optical phase conjugation with dispersion-flattened fiber (DFF) in WDM-PAM4 transmission systems. The proposed method increases the WDM channel to more than 16 in 26.6 Gbaud PAM4 data center interconnects.

Deep joint source-channel coding (DeepJSCC) is a promising technology for high efficiency wireless image transmission, but many existing models adopt a fixed compression rate. To address this problem, this paper proposes a DeepJSCC with lookup table generation network (LUT-GenNet), a novel scheme that provides peak signal-to-noise ratio (PSNR) guaranteed adaptive compression rate control. Conventional adaptive rate control approaches determine the compression ratio after encoding, which makes a processing bottleneck. Furthermore, since the encoder requires channel SNR information, if the channel SNR changes after it is obtained and encoding is performed, the image quality can degrade significantly. The proposed LUT-GenNet is a neural network that generates an image-specific LUT directly from the input image. This LUT instantaneously determines the optimal compression rate based on the target PSNR and channel SNR. This approach enables parallel processing of encoding and rate determination and avoids the SNR mismatch problem because the encoder does not rely on SNR feedback. A comparison with the conventional method under SNR mismatch conditions showed that while the conventional performance degrades significantly as the mismatch increases, the proposed method maintains stable.

Underwater acoustic communication (UWAC) faces significant challenges due to the severe multipath environment and limited bandwidth. Although orthogonal frequency division multiplexing (OFDM) is well known to attain high spectral efficiency, it requires long CPs to mitigate inter-symbol interference (ISI) due to existence of long multipath delay. This increases transmission overhead and undermines the advantages of OFDM. To address the above issue, this paper elucidates the impact of massive multiple-input multiple-output (MIMO) with shortened cyclic prefix (CP) frame design. An excessive spatial degree of freedom can auppress ISI that exceeds CP, enabling stable spatial multplexing transmission. Computer simulation using Bellhop, widely spread underwater propagation modeling tool, verifies its effectiveness.

This paper proposes a modeling method of semiconductor optical amplifier based on a deep learning technique. We report errors of gain and Q factor between the actual and estimated waveforms are less than 0.5 dB.

This paper proposes a modeling method of semiconductor optical amplifier based on a deep learning technique. We report errors of gain and Q factor between the actual and estimated waveforms are less than 0.5 dB.

Device-to-Device (D2D) communication is a promising solution for providing on-demand network connectivity to numerous devices. In particular, the autonomous D2D approach enables personal devices to flexibly communicate with each other with less operation. Despite all the benefits of D2D communication, security is a significant concern because of the broadcast nature of wireless communication. The biggest threats for the autonomous D2D are masquerading, impersonation, man-in-the-middle (MITM) attacks due to absence of a trusted third party. There have been many research efforts on this problem including physical layer based and the well-known Diffie-Hellman based approaches. However, they cannot be employed for authentication between physically distant devices. To address this problem, this paper proposes a multi-channel authentication for the autonomous D2D using optical camera communication (OCC). It executes the Diffie-Hellman key exchange in an optical link between a light source and a camera. The idea behind the proposed scheme is to leverage the limited reachability of OCC for ensuring security; a device can only communicate with a visible device. In this paper we introduce the security analysis for the proposed authentication and preliminary results using smartphones.

Optical eigenvalues originating in optical solitons are the potential for becoming information carriers not affected by chromatic dispersion and nonlinear effects in optical fibers. They are obtained by attributing the associated eigenvalue equations deduced by solving the nonlinear Schrödinger equation with inverse scattering transform (IST) to the matrix eigenvalue problem, and maintaining constant values regardless of the transmission distance. The eigenvalue communication systems require to solve the eigenvalues in soliton-by-soliton. While effective eigenvalue solution methods have not been studied well in telecommunication systems, one of the most well-known eigenvalue solution methods is the QZ decomposition-based method. However, the QZ algorithm requires a large complexity. To reduce the complexity, a method to demodulate the optical eigenvalues using a complex-moment eigenvalue solver (CME) was investigated. CME is a parallelizable eigenvalue solver that can extract any eigenvalue. This paper proposes a novel optical eigenvalue demodulation method that combines CME and an artificial neural network (ANN) based on employing an on-off encoded discrete eigenvalue modulation scheme. The ANN is sensitive to the input order of the units; therefore, the eigenvalues must be sorted. A lightweight sorting algorithm is hence required. In addition to the proposed scheme, this study proposes partial sorting using CME and ANN. Here, 2000 km fiber-transmission experiments for an on-off-encoded four-discrete-eigenvalue were conducted. The experimental results indicated that the proposed demodulation method obtained bit error rate (BER) characteristics comparable to conventional methods by devising an extraction range of eigenvalues in CME.

We propose a multi-stage adaptive equalizer for the demodulation of an optical aggregated 16QAM. A successful demodulation of the 100 Gbps class 16QAM signal was demonstrated without an OSNR penalty when the input PAM4 signals to the optical aggregation were not symbol-synchronized.

Ultra-low latency (ULL) data transmission is essential for many network applications. The IEEE 802.1Qbv Time-Aware Shaper (TAS) is a typical mechanism for providing ULL data transmission in bridged networks. All TAS enabled nodes must be time-synchronized and their gate time schedule in the gate control list (GCL) must be pre-configured considering the data arrival time and the maximum burst size. Consequently, the bandwidth utilization is drastically deteriorated if the data size becomes small, because no data can be forwarded until the state of the gate changes. Although the underutilization of link bandwidth can be addressed with gate shrunk (GS-)TAS which enables runtime reconfiguration of the GCL using a GS-frame at the end of an express traffic. However, the states of gates cannot be appropriately controlled with the existing scheme if there are many express streams. To address this issue, this paper proposes a concept of multi-stream gate control for TAS. The proposed idea enables runtime reconfiguration of the GCL to avoid the reduction in bandwidth utilization irrespective of the burst size and the number of streams. The performance of the proposed scheme was confirmed with computer simulations.

We propose a neural-demodulation-aided optimization algorithm for discrete eigenvalue assignment in nonlinear optical fiber systems. Using rank-based selection, our method achieves lower error rates than Monte Carlo sampling and experimentally demonstrates error-free transmission over 4000 km, surpassing conventional fixed-assignment eigenvalue modulation schemes.

This study investigates the validity of a nonlinear compensation based on optical phase conjugation (OPC) transmission with a dispersion-flattened fiber (DFF) in WDM-PAM4 transmission systems. We demonstrate that combining the OPC with the DFF enables suppression of the nonlinear distortion and an increase in maximum input power even for a high-speed transmission system, such as 106.3 Gbaud, through numerical simulations.

A passive optical network (PON) system is required to accommodate multi-services represented by fifth/sixth (5G/6G) mobile communication systems. While maintaining low cost, to meet these requirements (which are high capacity, low latency, and high reliability), the PON system must accommodate a large number of optical network units (ONUs). The use of power-domain non-orthogonal multiple access (PD-NOMA) for PON has been studied to increase the number of accommodatable ONUs. However, with an increase in the ratio of transmitted power between multiplexed ONUs, the effect of quantization noise in the analog-to-digital converter (ADC) is more significant. This study proposes to apply an optimal quantizer based on the Lloyd-Max algorithm (which is a nonlinear quantization algorithm) to PD-NOMA-PON. In general, the Lloyd-Max algorithm requires continuous probability density function when setting the partitions. We adopt a continuous function estimated from the discrete probability distribution extracted by the low-resolution ADC using an expectation maximization algorithm. We conduct experiments to verify the feasibility of the proposed quantizer for the downlink and confirm the improvement of the optical signal-to-noise ratio penalty for all power ratios.