In recent years, Optical neural Networks have received a lot of attention due to their high speed and low power consumption. However, these Networks still have many limitations. One of these limitations is implementing their nonlinear layer. In this paper, the implementation of nonlinear unit for an Optical convolutional neural network is investigated, so that using this nonlinear unit, we can realize an all-Optical convolutional neural network with the same accuracy as the electrical Networks, while providing higher speed and lower power consumption. In this regard, first of all, different methods of implementing Optical nonlinear unit are reviewed. Then, the impact of utilizing saturable absorber, as the nonlinear unit in different layers of CNN, on the network’, s accuracy is investigated, and finally, a new and simple method is proposed to preserve the accuracy of the Optical neural Networks utilizing saturable absorber as the nonlinear activating function.

This paper demonstrates the design of an Arrayed Waveguide Gratings (AWG) based Optical switch. In the design both physical and network layer analysis is performed. The physical layer power and noise analysis is done to obtain Bit Error Rate (BER). This has been found that at the higher bit rates, BER is not affected with number of buffer modules. Network layer analysis is done to obtain performance in terms of packet loss rate and average delay. Analysis presented in the paper clearly reveals that there is a minimum amount of power required, which is necessary for the satisfactory performance of switch both at physical and network layer.

The retina may be affected by many diseases such as Age-related Macular Degeneration (AMD), Diabetic Macular Disease (DME), and Choroidal Neovascularization (CNV). To diagnose these diseases, one way is to analyze retinal Optical Coherence Tomography (OCT) images using image processing algorithms. In this paper, a novel architecture based on Generative Adversarial Networks (GANs) and Convolutional Neural Networks (CNNs) is proposed to classify OCT images with insufficient training samples. The proposed method generates OCT images with pseudo labels from the distribution of original OCT images. OCT samples with real and pseudo labels are presented to a CNN classifier which leads to a model which is robust against insufficient samples. UCSD dataset has been used to evaluate the proposed method. Results indicate that the proposed method is comparable to the state-of-the-art methods in the terms of Precision, Sensitivity, Specificity, and F-Measure. Source code of this paper is available online at Github (https://github.com/mohamad-sw/oct-image-classification-using-gans).

In this essay, an Optical arrangement for the measurement of dynamic weight, by means of conducting a thin ray of light through it, is designed in order to enhance the measuring accuracy in the weight-estimation device. In this arrangement, CCD is responsible for producing raw data for processing. An artificial neural network type, RBF, is also used to improve the quality and speed of the measurement. While the scale of the weight-estimation device is oscillating, by applying the weight on the scale, the neural network by sampling the proportionate wave-shape yields the weight of object with high accuracy and high speed.

The aim of this paper is to fully analyze the effects of free space Optical (FSO) communication links on the estimation performance of the adaptive incremental Networks. The FSO links in this paper are described with two turbulence models namely the Log-normal and Gamma-Gamma distributions. In order to investigate the impact of these models we produced the link coefficients using these distributions and assumed that the network is exchanging data between the nodes that are contaminated with these coefficients. Firstly, by the FSO link assumption, we performed the theoretical analysis for the steady-state performance of the adaptive network and driven closed-form relations explaining the link impacts. Secondly, we performed simulation results for Log-normal and Gamma-Gamma link conditions and presented various results for different levels of turbulence. Finally, we compared the theoretical and analytical results showing a close agreement between these two findings. The results are presented by the means of mean square deviation (MSD) and excess mean square error (EMSE) values.

Nowadays, video semantic segmentation is used in many applications such as automatic driving, navigation systems, virtual reality systems, etc. In recent years, significant progress has been observed in semantic segmentation of images. Since the consecutive frames of a video must be processed with high speed, low latency, and in real time, using semantic image segmentation methods on individual video frames is not efficient. Therefore, semantic segmentation of video frames in real time and with appropriate accuracy is a challenging topic. In order to encounter the mentioned challenge, a video semantic segmentation framework has been introduced. In this method, the previous frames semantic segmentation has been used to increase speed and accuracy. For this manner we use the Optical flow (change of continuous frames) and a GRU deep neural network called ConvGRU. One of the GRU input is estimation of current frames semantic segmentation (resulting from a pre-trained convolutional neural network), and the other one is warping of previous frames semantic segmentation along the Optical flow. The proposed method has competitive results on accuracy and speed. This method achieves good performances on two challenging video semantic segmentation datasets, particularly 83. 1% mIoU on Cityscapes and 79. 8% mIoU on CamVid dataset. Meanwhile, in the proposed method, the semantic segmentation speed using a Tesla P4 GPU on the Cityscapes and Camvid datasets has reached 34 and 36. 3 fps, respectively.

The spectrally and spatially flexible Optical Networks (SS-FON) are the promising solution for future Optical transport Networks. The joint switching (J-Sw) paradigm is one of the possible switching schemes for SS-FON that brings Optical component integration alongside with acceptable networking performance. The network planning of JSw is investigated in this paper. The formulation of resource allocation for J-Sw is introduced as in integer linear programming to find the optimal solution. To find the near-optimal solution, the heuristic algorithms are initiated with sorted connection demands. The way connection demands are sorted to initiate the heuristic algorithms affects the accuracy of algorithms. Therefore, six different sorting policies are introduced for J-Sw. Moreover, the heuristic algorithm called joint switching resource allocation (JSRA) algorithm is introduced, especially for J-Sw. The heuristic algorithm performance initiated with different sorting policies is investigated through simulation for a small-size network. The optimality gap is the most important indicator that shows the effect of each sorting policy on the nearoptimal solution. The new sorting policy of connection demands called descending frequency width (DFW) policy achieved the least optimality gap. Also, the JSRA performance initiated with these sorting policies is investigated for a real network topology. The obtained results indicate that DFW shows better performance than other sorting policies in realistic Networks, too.

We address the throughput maximization problem for downlink transmission in DF-relay-assisted cognitive radio Networks (CRNs) based on simultaneous wireless information and power transfer (SWIPT) capability. In this envisioned network, multiple-input multiple-output (MIMO) relay and secondary user (SU) equipment are designed to handle both radio frequency (RF) signal energy harvesting and SWIPT functional tasks. Additionally, the cognitive base station (CBS) communicates with the SU only via the MIMO relay. Based on the considered network model, several combined constraints of the main problem complicate the solution. Therefore, in this paper, we apply heuristic guidelines within the convex optimization framework to handle this complexity. First, consider the problem of maximizing throughput on both sides of the relay separately. Second, each side progresses to solve the complex problem optimally by adopting strategies for solving sub-problems. Finally, these optimal solutions are synthesized by proposing a heuristic iterative power allocation algorithm that satisfies the combinatorial constraints with short convergence times. The performance of the optimal proposed algorithm (OPA) is evaluated against benchmark algorithms via numerical results on optimality, convergence time, constraints’ compliance, and imperfect channel state information (CSI) on the CBS-PU link.

All-Optical logic gates are the most important unit for achieving all-Optical processing systems. Developing a fast and efficient method for studying the behavior of all-Optical logic gates is very important and has been considered by researchers. In this paper, general regression neural Networks and linear method are used to predict a three-input all-Optical XOR logic gate output. The simulation results show that both methods can precisely model the behavior of the device. The training time of the neural network in the linear method with the optimal structure is about 93 seconds, which is much longer than the GRNN method with a training time of 8 seconds. Both models predict the output in less than 1 second which show a great improvement over the conventional method with 12 seconds. In the GRNN method with the smoothing factor of 0.001, the best results were obtained with MSE, RSE and MAE error values of 1.97×10-7, 5.95×10-6, and 1.6×10-4, respectively. In the linear method with 200 initial training data, the minimum values of MSE, RSE, and MAE are 1.11×10-22, 2.14×10-16 and 2.11×10-11, respectively, and the best modeled output is achieved. The value of correlation coefficient (R2) between the modeled output and the desired output of the logic gate is one for both neural network methods, which indicates a very good prediction for this method.

Due to the increasing growth of processing cores in complex computational systems, all the connection converted bottleneck for all systems. With the protection of progressing and constructing complex photonic connection on chip, Optical data transmission is the best choice for replacing with electrical interconnection for the reason of gathering connection with a high bandwidth and insertion loss on chip was mentioned. Optical routers play an important role in the Optical Network-on-Chip (ONoC), which are responsible for selecting the path between Optical signal source and the destination. In recent years, silicon Optical routers based on Micro-Ring Resonators (MRRs) and Mach-Zehnder Interferometers (MZIs) have been proposed. The design of Optical switches is desirable by using of Mach-Zehnder Interferometer. This is while that MRR Switches have low bandwidth, whereas Mach-Zehnder Interferometer switches have wide bandwidth inherently. Mach-Zehnder Interferometer switches are able to routing with high speed for data transmission with Nano second switching time. This is while, that MRR switches in compare to MZIs has the less power consumption and area consumption. On the other hand we can divide Optical routers into parts, A. general router and B. specific-router, so that in specific routers, some of I/O paths for the reason of avoiding deadlock had be omitted. In continue, several kinds of Optical router based on MZI and MRR along with researching a series of parameters was mentioned.