A document by Babak Tavassoli . Click on the document to view its contents.

In this paper, we focus on simple data-driven approach to solve deep learning based on implementing the Mask R-CNN module by analyzing deeper manipulation of datasets. We firstly approach to affine transformation and projective representation to data augmentation analysis in order to increasing large-scale data manually based on the state-of-the-art in views of computer vision. Then we evaluate our method concretely by connection our datasets by visualization data and completely in testing to many methods to understand intelligent data analysis in object detection and segmentation by using more than 5000 image according to many similar objects. As far as, it illustrated efficiency of small applications such as food recognition, grasp and manipulation in robotics

We report a broadband radio frequency (RF) channelizer with up to 92 channels using a coherent microcomb source. A soliton crystal microcomb, generated by a 49 GHz micro-ring resonator (MRR), is used as a multi-wavelength source. Due to its ultra-low comb spacing, up to 92 wavelengths are available in the C band, yielding a broad operation bandwidth. Another high-Q MRR is employed as a passive optical periodic filter to slice the RF spectrum with a high resolution of 121.4 MHz. We experimentally achieve an instantaneous RF operation bandwidth of 8.08 GHz and verify RF channelization up to 17.55 GHz via thermal tuning. Our approach is a significant step towards the monolithically integrated photonic RF receivers with reduced complexity, size, and unprecedented performance, which is important for wide RF applications ranging from broadband analog signal processing to digital-compatible signal detection.

In this paper, we investigate mid-IR SC generation in a cascaded silica-ZBLAN-chalcogenide fiber system directly pumped with a commercially-available 460-ps pulsed fiber laser operating in the telecommunications window at 1.55 µm. This fiber-based system is shown to generate a flat broadband mid-IR SC covering the entire range from 2 to 10 µm with tens of mW of output power. This technique paves the way for low cost, practical, and robust broadband SC sources in the mid-IR without the requirement of mid-infrared pump sources or Thulium-doped fiber amplifiers. We also describe a fully realistic numerical model used to simulate the nonlinear pulse propagation through the cascaded fiber system and use our numerical results to discuss the physical processes underlying the spectral broadening in the cascaded system. We conclude with recommendations to optimize the current cascaded systems based on the simulation results.

In Internet of Things (IoT) applications, among various authentication techniques, keystroke authentication methods based on a user’s touch behavior have received increasing attention, due to their unique benefits. In this paper, we present a technique for obtaining high user authentication accuracy by utilizing a user’s touch time and force information, which are obtained from an assembled piezoelectric touch panel. After combining artificial neural networks with the user’s touch features, an equal error rate (EER) of 1.09% is achieved, and hence advancing the development of security techniques in the field of IoT.

Concurrent high force detection accuracy and extended battery lifetime are expected for wearable gait monitoring systems. In this article, a piezoelectric insole device and rectifying circuitry-based technique is presented to achieve these two goals. Here, walking induced positive and negative charges are separated for plantar stress detection and energy harvesting respectively, realizing the two functions concurrently. The high detection sensitivity of 55 mN and responsivity of 231 mV/N are achieved, satisfying the need for diagnosing various diseases. 1.6 pJ is stored during a walking event, extending the battery lifetime. The developed technique enhances the development of gait monitoring in IoHT.

In this article, we propose a new sensor architecture, in which electrodes are settled at the edge of the touch panel, ensuring a very high optical transmittance. The touch event detection relies on electrical capacitance tomography (ECT) technique. The presented technique provides a feasible means to boost the optical transmittance of the touch panel layer while maintaining touch detection accuracy.

Power traders and system operators need to balance the uncertain generation of renewable energy sources by adapting the dispatch of conventional power plants. This poses a challenge to voltage control in power system operation. Accordingly, this paper presents a method to determine voltage magnitude probability densities which are integrated into an optimal reactive power flow to consider an uncertain active power generation. First, the probability densities are determined by Monte Carlo simulations including a unit commitment problem to derive the dispatch of conventional power plants. Second, uncertainty restrictions are used to create soft constraints for the optimal reactive power flow, which mitigates the risk of voltage limit violations. By adapting the soft constraints slack costs, the consideration of uncertainties can be prioritized in relation to other objectives, such as the reduction of active power losses, or reactive power costs.

At present, all superconductivity techniques are carried out at ultra-low temperature. The aim of this study is to design a novel vacuum tube at room temperature, which can move electrons without resistance to achieve zero resistance and superconductivity properties. superconductor; vacuum tube; electronics; electrostatic shielding; zero resistance

The energy system decarbonization and decentralization require coordination schemes for distributed generators and flexibilities. One coordination approach is local energy markets for trading energy among local producers and consumers. The resulting local coordination leads to the questions of how the interaction between local and wholesale markets will be designed and of how the introduction of local energy markets influences the wholesale market system. Therefore, this paper proposes a bottom-up modeling method for local markets within a pan- European wholesale market model. Furthermore, an aggregation-disaggregation method for local markets is developed to reduce computational effort. A case study for local markets in Germany shows the computational advantages of the aggregation-disaggregation method. Preliminary results indicate the impact of different interaction designs between local and wholesale markets on the wholesale market and show the need for further research.

Abstract: To enable full-duplex (FD) in underwater acoustic (UWA) systems, a high level of self-interference (SI) cancellation (SIC) is required. For digital SIC, adaptive filters are used. In time-invariant channels, the SI can be effectively cancelled by classical recursive least-square (RLS) adaptive filters, such as the sliding-window RLS (SRLS) or exponential-window RLS, but their SIC performance degrades in time-varying channels, e.g., in channels with a moving sea surface. Their performance can be improved by delaying the filter inputs. This delay, however, makes the mean squared error (MSE) unsuitable for measuring the SIC performance. In this paper, we propose a new evaluation metric, the SIC factor (SICF), which gives better indication of the SIC performance compared to MSE. The SICF can be used in experiments and in real FD systems. A new SRLS adaptive filter based on parabolic approximation of the channel variation in time, named SRLS-P, is also proposed. The SIC performance of the SRLS-P adaptive filter and classical RLS algorithms (with and without the delay) is evaluated by simulation and in lake experiments. The results show that the SRLS-P adaptive filter significantly improves the SIC performance, compared to the classical RLS adaptive filters.

Digital image steganalysis is the process of detecting if an image contains concealed data embedded within its pixel space inserted via a steganography algorithm. The detection of these images is highly motivated by Advanced Persistent Threat (APT) groups, such as APT37 Reaper, commonly utilizing these techniques to transmit malicious shellcode to perform further post-exploitation activity on a compromised host. Performing detection has become increasingly difficult due to modern steganography algorithms advancing at a greater rate than the steganalysis techniques designed to combat them. The task of detection is challenging due to modern steganography techniques that embed messages into images with only minor modifications to the original content which varies from image to image. In this paper, we pipeline Spatial Rich Models (SRM) feature extraction, Principal Component Analysis (PCA), and Deep Neural Networks (DNNs) to perform image steganalysis. Our proposed model, Neural Spatial Rich Models (NSRM) is an ensemble of DNN classifiers trained to detect 4 different state-of-the-art steganography algorithms at 5 different embedding rates, allowing for an end-to-end model which can be more easily deployed at scale. Additionally our results show our proposed model outperforms other current state-of-the-art neural network based image steganalysis techniques. Lastly, we provide an analysis of the current academic steganalysis benchmark dataset, BOSSBase, as well as performance of detection of steganography in various file formats with the hope of moving image steganalysis algorithms towards the point they can be utilized in actual industry applications.

Edge processing for computer vision systems enable incorporating visual intelligence to mobile robotics platforms. Demand for low power, low cost and small form factor devices are on the rise.This work proposes a unified platform to generate deep learning models compatible on edge devices from Intel, NVIDIA and XaLogic. The platform enables users to create custom data annotations,train neural networks and generate edge compatible inference models. As a testimony to the tools ease of use and flexibility, we explore two use cases --- vision powered prosthetic hand and drone vision. Neural network models for these use cases will be built using the proposed pipeline and will be open-sourced. Online and offline versions of the tool and corresponding inference modules for edge devices will also be made public for users to create custom computer vision use cases.

This paper presents novel designs of frequency reconfigurable distributed non-reciprocal bandpass filter and diplexer based on spatio-temporally modulated microstrip λg/2 resonators. The modulation is achieved by loading both ends of the λg/2 transmission line resonators with time-modulated capacitors. To provide an inherent biasing isolation between the RF and the modulation signals, the modulation voltage source is connected at the center of the resonator, where there is a natural voltage null. A single inductor is used to further enhance such biasing isolation. The wideband nature of this isolation scheme enables the tuning of the devices over a wide frequency range. With more than 30-dB RF to modulation isolation, the proposed resonator structure also enables low insertion loss by eliminating RF signal leakage to the modulation ports. Two examples of a 3-pole bandpass filter and a diplexer are demonstrated with good agreement between the measurement and the simulation. The fabricated filter shows a minimal insertion loss of 3.9 dB, a 20-dB isolation bandwidth of 42 MHz at 1.0 GHz, and frequency tuning range of 885-1031 MHz. The measured diplexer has two non-reciprocal bandpass channels at 829 MHz and 997 MHz, respectively. The two channels can be independently reconfigured without affecting each other.

To improve the state estimation accuracy of nonlinear induction motor with uncertain parameters, a robust desensitized rank Kalman filtering (DRKF) is proposed to reduce state estimation error sensitivities to uncertain parameters. A new sensitivity function is defined, and a novel desensitized cost function for the deterministic sampling methods is designed to obtain an optimal gain matrix. The sensitivity propagation is summarized for deterministic sampling methods. Based on the rank sample rule, the sensitivity propagation method is given, and the DRKF algorithm is derived. Two dynamic behaviors of the induction motor with two uncertain stator and rotor resistances are simulated to demonstrate that the proposed DRKF has an excellent performance.

This manuscript was submitted to IEEE Access on 12 Jun 2020. Abstract: Simulation forms an important part of the development and empirical evaluation of underwater acoustic network (UAN) protocols. The key feature of a credible network simulation model is a realistic channel model. A common approach to simulating realistic underwater acoustic (UWA) channels is by using specialised beam tracing software such as BELLHOP. However, BELLHOP and similar modeling software typically require knowledge of ocean acoustics and a substantial programming effort from UAN protocol designers to integrate it into their research. In this paper, we bridge the gap between low level channel modeling via beam tracing and automated channel modeling, e.g. via the World Ocean Simulation System (WOSS), by providing a distilled UWA channel modeling tutorial from the network protocol design point of view. The tutorial is accompanied by our MATLAB simulation code that interfaces with BELLHOP to produce channel data for UAN simulations. As part of the tutorial, we describe two methods of incorporating such channel data into network simulations, including a case study for each of them: 1) directly importing the data as a look-up table, 2) using the data to create a statistical channel model. The primary aim of this paper is to provide a useful learning resource and modeling tool for UAN protocol researchers. Initial insights into the UAN protocol design and performance provided by the statistical channel modeling approach presented in this paper demonstrate its potential as a powerful modeling tool for future UAN research.

This paper presents a new combination of Decomposition Filters (DF) and Hyperbolic Tangent (HBT) based simplified directional operators for edge detection (ED). Conventional separable algorithms are limited in capturing the geometric features and non-separable filters eliminate these restrictions. The high-frequency band of the DF of proposed method achieves the edge information. A set of edge operators in the form of simple non-separable patterns for different scales, and orientations are applied on the edge information to capture linear edge structures and directions. While testing with ten natural images, this algorithm has improved performances in terms of reduced Mean Squared Error (MSE) in the reconstruction estimation measure. The performance measure compares the Simplified Gabor Wavelet (SGW) based ED.

This paper deals with a cloud radio access network (CRAN) architecture for the LoRa system. In the suggested design, the gateway embeds a limited remote radio head (RRH), including the analog radio-frequency (RF) analog part, the digital-to-analog and analog-to-digital conversion, and a digital front-end (DFE). The other LoRa network functions, including the physical (PHY) layer, the LoRaWAN medium access control (MAC) layer, and the application and customer servers are implemented as cloud resources. The presented approach leads to a flexible RAN that is robust to the variations of capacity needs. Furthermore, it allows us to test very specific LoRa features, such as the detection or demodulation, while bypassing the other ones including the hardware RRH. The methodology and tools we used to deploy a LoRa cloud RAN are detailed, and results concerning the performance indicator (CPU load, memory consumption) are provided as well.