Iraqi Journal of Information and communication technology

Recently, the DVB-C standard is used to perform the function of transmitting digital television signalsthrough communication cables and transmitting signals with the ability to maintain the speed and reliability ofdigital video streams as well as audio streams to be displayed online via television without delay and with anacceptable low BER. In this article, the DVB-C standard was used to perform the same functions mentioned aboveby proposing a model using MPEG-4 streams of digital video and image from the transmitter with the use of QAM64modulation technology with reliability assurance to meet the requirements of the future in television broadcasting.Communication system toolkit in MATLAB Simulink is utilized to implement the proposed model. In order tobuild the channel for the proposed system, the Adaptive White Gaussian Noise (AWGN) channel is utilized. Thepower spectrum density is calculated, which were acceptable with the guarantee of no delay. Also, the resolutionis enhanced in rate of 98%. Also, the input signal and the output signal are compared, and the semi completematch has obtained. The proposed model was created using the MATLAB 2023B simulation and program packages.

Nowadays, electronic olfaction or Electronic Noses (E-Noses) are mostly designed and developed forlab-based applications. In real-life scenarios, smell detection represents the external and internal responses tohuman being events. In Human Computer Interaction (HCI), detecting odorant molecules in the air plays a veryimportant role in different conditions such as food quality, environmental monitoring, and security. Few real-timeapplications based on E-Nose have been developed, such as biosensors based on insect antenna or small drones withlow-cost gas sensors. Smell detection and recognition as an environmental monitoring application using sensor arraymeasurements represent the scope of this work. To increase the accuracy and application range of detection, sixof high and low-quality sensors have been used to get measurements. In this study, the proposed E-Nose prototypeconsists of DF-NH3, MQ-136, MQ-135, MQ-8, MQ-4, and MQ-2 sensors. This work presents the methodologyin terms of artificial intelligence techniques used for classification and object detection such as machine learningclassification, clustering and regression algorithms. Using the K-Nearest Neighbors (KNN) algorithm, these sensorsused in the experimental results were able to predict petrol, gasoline, perfumes, and others with a 71.1% accuracyrate

Enhanced YOLOv8 for Edge computing with a deep learning approach for memory optimization isproposed as an optimized Artificial Intelligence (AI) algorithm specifically designed for edge learning, which canefficiently enable Convolutional Neural Networks (CNN) deployment on Internet of Things (IoT) edge devices withless memory footprint and more rapid inference. This paper optimizes the YOLOv8 model architecture using thestructured pruning method (Layer Pruning) to minimize non-contributing parameters. The proposed pruned modelwith loaded pre-trained model weights on the COCO dataset and after fine-tuning on the self-driving car datasetmaintains high accuracy, with a negligible 0.011 decrease compared to the baseline, while reducing the model sizeby 33% and increasing detection speed to 196 Frame Per Second (FPS). These results are significant in making theproposed model well-suited for edge environments with limited memory capacity, especially for the dataset usedallowing the vehicle to react rapidly to environmental changes for practical applications.

Cloud computing is defined as the practice of storing, managing, and processing data on a networkof remote servers hosted on the internet rather than a local server or a personal computer. Cloud computing isamong the newest technologies in era of Information Technology (IT). Most firms employ cloud computing to savemoney on server, storage, and maintenance infrastructure purchases. Addressing the several problems related tocloud computing in depth is therefore rather crucial. A big problem for industries is security. In this paper, weinvestigate some major advantages of cloud computing and significant features of cloud model. The report alsocovers security concerns, drawbacks of the current method, and possible solutions related to cloud computing.

In the evolving landscape of quantum computing, the security of conventional cryptosystems is increas-ingly compromised. This requires innovative data protection strategies, leading to the exploration of Post-QuantumCryptography (PQC) techniques. This research aims to address these challenges, focusing on the integration of Nth-Degree Truncated Polynomial Ring (NTRU) encryption - known for its quantum resistivity - and Shamir’s SecretSharing (SSS) for secure and quantum-secured key distribution, particularly in resource-constrained environments.Despite their effectiveness, NTRU ciphers face obstacles in public key distribution due to significant computationaland memory requirements. Leveraging SSS, a scheme to distribute the NTRU public key to multiple shares, providingenhanced security while mitigating resource limitations is proposed. Recognizing the potential for security breacheslike Man-in-the-Middle (MitM) and Distributed Denial of Service (DDoS) attacks, the proposed work seeks to fortifydefenses against these threats. The findings of the proposed method are intended to offer a significant contributionto strengthening cryptographic practices in the face of emerging quantum technologies.

In the environment of multiple edges, an unbalanced distribution of offloaded tasks can result in alack of edge resources, which in turn which in turn leads to lower performance. On the other hand, fast decisionsregarding edge selection are crucial for efficient performance. Therefore, this paper suggests an edge-edge networkbased on software-defined networks to manage resources and tasks at the mobile edge of computing in the Internetof Things (IoT) environment. The proposed technique in this paper introduces an effective method for makingselections concerning collaborative offloading tasks within edge computing environments based on Software-DefinedNetworks (SDN), which will decide where to offload and process tasks on the optimal Mobile Edge Computing(MEC) server among five MEC servers based on currently available resources when tasks need processing duringa specific time using SDN controller that view the status of all network. The rank of the feasible MEC server isbased on the presently available CPU frequency of the MEC server relative to the required computing resourcesfor the task. To calculate the final height score of the MEC server, this work used Min-Max normalization and ahigh score for the MEC server from these servers that were considered optimal for offloading tasks. This paperaims to maintain the total latency as little as much as possible.

Recent advances in genetics have increased the sensitivity and reliability of the forensic sciences.Therefore, there is a need for efficient forensic investigation techniques. Deep learning is becoming increasinglyimportant in forensic science as it offers the potential to increase the accuracy and effectiveness of various forensictasks like paternity testing, missing person identification, and potentially connecting suspects to crime sites. Althoughit’s performance at solving many problems, the deep learning model may suffer from overfitting problems thatoccur when it fails to generalize well and instead fits more precisely to the training dataset. This work presentstwo deep learning models, Deep Neural Network (D-DNN) and Gated Recurrent Neural Network (D-GRU), forhuman identification based on Deoxyribonucleic Acid-Short Tandem Repeat (DNA-STR) as the input sequence.These models are built and tested in such a way as the best performance since two regularization techniques areintroduced: dropout and data augmentation to avoid overfitting problems. Two datasets are used: one with a size of53530 and another with a size of 151580. Whereas 80% for training purposes of the first dataset is equal to 42824,while it is equal to 121264 in the second dataset. A comparison of performance was held between these two modelsby using dropout or not. The results show that D-GRU using dropout has the best performance with overfittinginvisibility, training and testing accuracy equal to 1.0, the loss equal to 1.5276×10−8, and the validation loss equalto 6.1561×10−6.