فصلنامه پدافند الکترونیکی و سایبری
سال یازدهم شماره 3 (پیاپی 43، پاییز 1402)

With the advent of quantum computing theory and quantum communication networks, establishing confidential and secure communication has become challenging. Quantum audio signal steganalysis is one of the interesting subfields in the field of quantum signal processing and quantum computing, which tries to identify hidden communications in the context of quantum communication networks by using feature extraction techniques and quantum machine learning algorithms. Since steganography causes inevitable changes in the statistical characteristics of the frequency domain of the host signal, it can be used as an efficient and effective tool to build comprehensive and accurate steganalysis. So; At first, the proposed method uses quantum Fourier transform on QRDS audio signal to extract statistical features. For this purpose, the proposed quantum circuit network of these features, quantum spectral center and quantum spectral bandwidth has been designed and implemented. Finally, the Quantum Support Vector Machine (QSVM) algorithm, using the extracted features, separates clean and stego data sets with an accuracy of more than 95%.

Tor is one of the most popular networks providing anonymity and privacy on the Internet that works using volunteer systems from around the world. Low latency operation makes it suitable for things like web browsing. The way to select a path in tor network is one of the influential factors in the efficiency and security of this network. The path selection algorithm in the tor network has undergone many changes over its original design. These changes have been made for reasons such as increased efficiency, reliability, and load balancing, sometimes even in response to introduced attacks, they have to change the path selection algorithm. So far, many papers have looked at how to choose relays in the creation of circuits, but none of them, despite the open-source of the tor code, did not analyze the routing algorithm and the weighting method of the relays. And they have not proved logically and mathematically the relations used in this algorithm. In this paper, we attempt after fully analyzing the tor routing algorithm, for the first time, the exact logical and mathematical proofs of the relationships used in this algorithm are discussed. In this paper, we tried to investigate the anonymous network routing algorithm as the largest current anonymous network in the world.

In underwater acoustic channels, due to the limited bandwidth, the extension of the long delay due to the low speed of sound propagation, and also due to severe time changes, the establishment of a stable and efficient communication has always been accompanied by obstacles. Therefore, the use of orthogonal multi-input multi-Output multi-Carrier systems is common, but the challenge is that due to the use of long-distance rotary prefix for underwater channels, bandwidth gain and data transfer rate are greatly reduced. The proposed solution in this paper is to use multi-carrier systems based on offset bank filters. By using these systems, due to the lack of rotational prefix, the orthogonal frequency division system will no longer have problems and therefore will see an increase in interest and an increase in transmission rate in its multi-input-multi-output type. In this article, for the first time, a Hermit pulse forming filter is used, which has more accuracy and operational power than other filters. Also, in order to eliminate the destructive interference, the equalizer of at least average squares error has been used. The simulation results show the optimal performance of the proposed subsurface transmission system in shallow water. The output of simulations and numerical analysis has shown that at the rate equal to the error rate, the proposed system has a 15% higher send and receive rate, which is a valuable achievement in complex underwater conditions.

Today, strategic UAVs with their ultra-modern facilities in the fields of COMMAND, ELINT and high-resolution imaging and their guidance and navigation and sending and receiving information at a high rate through communication links with satellites are able to collect valuable information. which can disturb the balance of the battle scene and if needed, they will be able to act and destroy the country's strategic infrastructures. Therefore, in order to deal with this type of birds, it is felt to use an electronic warfare system that is capable of active and passive tracking with the ability to disrupt. In this system, due to the many advantages of antenna array and electronic beam shaping, this method has been used. Depending on the operational needs, these arrays should be capable of receiving several Mhz instant bandwidth signal over the entire operating frequency range x and Ku. In array antennas by sending and receiving broadband signals, the narrow beamforming structure will not respond to the beamforming, so wideband beamforming structures are used. In broadband structures, due to the increase in the number of beamforming coefficients from the M coefficients to the M * J coefficients, if the common optimization algorithms in wideband beam formation are used to determine the coefficients, computational complexity and therefore power The required processing and computational latency are very high, which is one of the challenges of beamforming in wideband systems. In this paper, in order to reduce the computational complexity, the deep learning method has been used and it is shown that the proposed method reduces the complexity by determining the coefficients significant while maintaining efficiency.

With the increase in the Internet's penetration rate in life and the use of this technology in all aspects, the use of mobile phones has increased as well. This, in addition to creating many benefits, has expanded and accelerated the release of some malicious programs called malware. In this study, it is attempted to use a multilayer neural network and learning machine diagnosis of zero daytime malware on smartphones. For this purpose, the standard database has been labeled with more than 15,000 samples of malware and goodware. In the pre -processing phase, the data is first performed using normalization and alignment of the data and by analyzing the main components of the feature of the selection of the feature and selected from 1183 features 215 features that have higher variances, followed by the model. A suggestion is introduced from the multilayer neural network class and the optimization algorithm based on the training and learning that apply it to the databases and compare its classification results with vector algorithms, genetic algorithm, nearest neighbor. And ... it can be seen that the neural network training increases accuracy and accuracy. The results of the use of multilayer neural network based on education and learning indicate 99% accuracy and 98% accuracy.

Classification and recognition is one of the most important methods of extracting information from images, and among them, facial image recognition as one of the most efficient biometric features for human identification has always been of interest, and extensive research has been conducted in this field in recent years. So far, various solutions for face recognition have been proposed by researchers, but among them, the use of Sparse representation classification has been considered as an effective and specific solution. One of the features of Sparse representation is to obtain features from input images without the need of feature extraction methods, therefore, in this article, the proposed method is aimed at applying weighted smoothed ℓ0 norm for face recognition using Sparse representation.To check the performance of the proposed method, ORL and AR databases including images of different facial expressions have been used, and the simulated results show that the method performs very well compared to other well-known methods in the field of face recognition.

Software engineers are always looking to reduce production costs and increase software quality. There are various methods to improve software quality, and code refactoring is one of these methods. Code refactoring and reorganization is a method for cleaning up software code and is one of the crucial processes in maintaining software quality. One of the main challenges in developing and producing clean code in software is the existence of inconsistent or bad-smelling code. Code smell is a superficial sign in the code that may indicate a deeper problem in the software. The existence of code smells may slow down processing, increase the risk of failure, as well as software errors. Therefore, software developers attempt to identify inconsistent code and facilitate its maintainability and scalability by refactoring software code. However, manual and automatic identification of code smells is challenging and tiring. As a result, methods for identifying such codes automatically and semi-automatically have been proposed. An important note in non-automatic methods is that predicting inconsistent code requires individual knowledge that is both time-consuming and increases the possibility of error. Therefore, automated methods have a greater advantage in predicting inconsistent code. So far, extensive research has been conducted on automatic prediction and identification of inconsistent code. A high percentage of these studies have focused on predicting four types of code smells: long method, feature envy, god class, and data class. In this article, our focus is on improving the accuracy of extracting such inconsistent codes. One of the common methods for predicting this type of code is using machine learning-based methods. Artificial neural networks are a specific type of machine learning algorithm that is modeled according to the human brain's performance method. This means that these networks can learn from input data and provide responses in the form of predictions and classifications. In this article, a multi-layer neural network was used to predict software inconsistent code, as well as a new feature selection method to increase prediction accuracy.

In the IoT-based IoT environment, users can monitor tasks in the network environment and participate in the data collection process by smart devices. Users monitor their environment data in the form of fog computing during this process, also called mobile mass monitoring, and service providers are required to pay user rewards. But rewards should not be such as to increase platform costs. At the same time, maximizing the maximization rate is one of the main goals of service providers. Increasing network coverage rates and reducing platform costs can be considered as an optimization problem. But providing an algorithm that is less involved in local optimizations and can always provide good results is a challenge in itself. This article is tried to present an efficient approach based on the improved forest optimization algorithm using chaos theory and fuzzy parameter adjustment to reduce platform costs and maximize mobile mass monitoring coverage rate. The proposed method is implemented in MATLAB software and the analysis of the findings shows that the proposed method can optimize the network coverage rate by 31% (average) and the monitoring cost by 11% (average) compared to the CMST plan.

In the field of electronic warfare, acquiring information, detecting, localization, and tracking the target is one of the most important topics in the support sector. Concealment and anti-jamming are significant needs in this sector. This research, by designing and analyzing the radiation-free structure of optical fiber sensing, took advantage of the property of the formation of speckle light spots in the output of the multi-mode optical fiber, and in addition to detecting the deformation of the optical fiber, which can be due to pressure or displacement, By processing the output signal of a photodiode placed at a suitable distance from the end of the fiber, the dynamic changes caused by mechanical vibration was detected and measured. In addition to the frequency, the vibration amplitude is also extracted. In this research, dynamic change was applied to a part of the optical fiber length by using a shaker drive device that can be controlled in terms of amplitude and frequency, and the signal output after amplification and filtering was detected and identified. The frequency range from 5 to 91 Hz and the displacement range of the fiber from 50 μm to above were tested. The frequency and amplitude of the shaker correspond to the output electrical signal, and the sensitivity was studied based on the fiber Specklegram. The detection and extraction of the frequency parameter of the mechanical stimulus and the displacement of the fiber through the processing of the speckle spots at the end of the optical fiber along with the sensitivity measurement towards the amplitude and low frequency are reported in this article.

Distributed Denial of Service (DDoS) attack is the widespread sending of valid or invalid packets to a server on the Internet, occupying its bandwidth and preventing execute legitimate requests of other users. The best approach to secure the network from such attacks is to exploit security controls such as intrusion detection and prevention systems. Cyber security researchers have significantly focused on identifying and counteracting this attack and have increased the accuracy and performance of security systems by providing various artificial intelligence solutions. The purpose of this paper is also to provide a solution for detecting DDoS attack, where, decision tree, multi-layer perceptron and random forest classifiers have been utilized in an ensemble method to mitigate the over-fitting problem. Also, two approache, i.e., batch learning and active learning have been implemented and evaluated in the classification phase of the proposed method. The evaluation results show that the mean value of accuracy in DDoS attack detection is 99.81%.