In this study, the sequential maximum likelihood decoding (S-MLD) for image sensor based visible light communication, referred to as image sensor communication, is proposed. The conventional system cannot correctly decode degraded captured images. To overcome this problem, S-MLD partitions the LED area on the image into two segments and sequentially creates pseudo-captured images, which imitate all possible LED blinking patterns. This approach is based on MLD for each segment. Moreover, reliability determination is incorporated to improve communication quality. We performed computer simulations and confirmed that S-MLD exhibits superior decoding performance and approximately 90% computational cost reduction compared to the previous MLD.
Recently, various disasters such as terrorism as well as natural disasters have occurred. We have developed the evacuation support system named as Emergency Rescue Evacuation Support System (ERESS) to reduce the victims. This network system consists of mobile terminals, recognizes disasters through various sensors, and presents appropriate evacuation routes. In this paper, we propose a method to recognize disasters immediately and accurately by using Support Vector Machine (SVM) and location information. This method uses two different SVM algorithms that are Buffering and Bagging-SVM to analyze owner behavior and recognize disasters. We show the validity of the proposed method by panic-type experiments and simulations.
We propose a novel sextuple-frequency two-tone signal generation method using two cascaded Mach-Zehnder modulators (MZMs). The first MZM is used to generate a large third-order harmonics component with a large RF input of 4.64Vπ (peak-to-peak). The second MZM is used to suppress first- and fifth-order unwanted harmonics at the same time. The proposed method was investigated by numerical simulations and experiments. We successfully suppressed the unwanted harmonics to less than -21 dB of the intended third-order harmonics by using two MZMs with extinction ratios of 23 dB and 26 dB.
This study investigates pump light generation methods for all-optical wavelength conversion systems using four-wave mixing in highly nonlinear fibers. The number of light sources, combination of sinusoidal signal frequency, and number of cascaded stages of LiNbO3 phase modulators were compared. As a result, 13 flat optical combs with a 10-GHz frequency spacing were generated with three-stage modulators, and it was confirmed that the noise level of the pump was as small as the seed light source. In the wavelength conversion experiment, a bit error rate of <10-10 for the converted signals was obtained in the case of the ±60-GHz shift from the original carrier pump lights.
In this letter, we propose a location estimation method for an unknown signal source. We considered this problem as nonlinear optimization one and found an effective solution. First, RSSI samples associated with location information are obtained. This problem is solved by finding the position and effective radiated power at which the error between the measured RSSI and the RSSI estimated from an assumed propagation loss equation is minimized. The validity of this method was clarified by numerical evaluation.
We numerically assess the performance and computational complexity of split-step Fourier method-based digital backpropagation with optical-field-intensity averaging (FIA-DBP) for reducing digital signal processing complexity in an optical receiver. We propose FIA-DBP to effectively compensate for nonlinear distortions due to symbol interactions. FIA-DBP averages the optical field intensity estimated in every span from received signals. We show the numerical results for 56Gbaud single carrier dual polarization-64 quadrature amplitude modulation signals. Our proposed method achieves a complexity 50.8% less than that of conventional digital backpropagation at a transmission distance of 2400km and a launched power of -2dBm.