No abstract

Despite the development of vaccines and the emergence of various treatments for COVID-19, the number of confirmed cases of the coronavirus disease (COVID-19) is increasing worldwide, and it is unlikely that the disease will ever disappear completely. Having a non-contact remote testing system can improve the workload of health-care centers and contribute to reducing the infection by recommending early self-isolation for those who suffer from a cough. In the proposed system, patients can upload an audio cough recording via mobile phones through the suggested Cough/X-ray/CT website and then receive the diagnosis within seconds on the same phone. Moreover, in the case of infection, the health center and the community are informed in addition to automatically calling the mobile phones of the injured cases. The higher proposed accuracy with deep cough training was achieved on the ResNet152v2 model after converting the cough signal into an image using the Mel-spectrogram, where the accuracy was 99.95%, the sensitivity was 100%, and the specificity was 99%.

It is no secret to all that the corona pandemic has caused a decline in all aspects of the world. Therefore, offering an accurate automatic diagnostic system is very important. This paper proposed an accurate COVID-19 system by testing various deep learning models for x-ray/computed tomography (CT) medical images. A deep preprocessing procedure was done with two filters and segmentation to increase classification results. According to the results obtained, 99.94% of accuracy, 98.70% of sensitivity, and 100% of specificity scores were obtained by the Xception model in the x-ray dataset and the InceptionV3 model for CT scan images. The compared results have demonstrated that the proposed model is proven to be more successful than the deep learning algorithms in previous studies. Moreover, it has the ability to automatically notify the examination results to the patients, the health authority, and the community after taking any x-ray or CT images.

Today, microcontrollers are of paramount importance in various aspects of life. They are used for design in many industrial fields from simple to highly complex devices. With a COVID-19 crisis going on, blending learning is the ideal solution for a post-pandemic society. This paper proposes a blended learning system as a solution to address today's problem in teaching microcontroller courses through collaboration between distance learning with the proposed training toolkit for real work. Implementation of the proposed solution began by constructing an inexpensive training kit (100$), to empower all students, even those in remote rural areas. The distance learning model allows the simulation of the proposed IoT projects electronically anywhere and at any time using the Proteus design suite, which helps students to conduct them before the actual laboratory appointment. Two learning models are programmed in assembly language which is directly related to the internal architecture of the microcontroller and provides access to all the real capabilities of its central processing unit. To get acquainted with all the features offered by the microcontroller integrated circuit, various IoT projects were constructed, each one dedicated to learning its architecture features, important to engineering students. The proposed IoT systems operate with a minimum consuming power that is very important for portable devices.Questionnaire questions for students were formulated to measure the proposed system benefit over three academic years.

Steganography is the idea of hiding secret message in multimedia cover which will be transmitted through the Internet. The cover carriers can be image, video, sound or text data. This paper presents an implementation of color image steganographic system on Field Programmable Gate Array and the information hiding/extracting techniques in various images. The proposed algorithm is based on merge between the idea from the random pixel manipulation methods and the Least Significant Bit (LSB) matching of Steganography embedding and extracting method.In a proposed steganography hardware approach, Linear Feedback Shift Register (LFSR) method has been used in stego architecture to hide the information in the image. The LFSRs are utilized in this approach as address generators. Different LFSR arrangements using different connection unit have been implemented at the hardware level for hiding/extracting the secret data. Multilayer embedding is implemented in parallel manner with a three-stage pipeline on FPGA.This work showed attractive results especially in the high throughputs, better stego-image quality, requires little calculation and less utilization of FPGA area. The imperceptibility of the technique combined with high payload, robustness of embedded data and accurate data retrieval renders the proposed Steganography system is suitable for covert communication and secures data transmission applications.