JOURNAL OF ELECTRONIC AND CYBER DEFENCE
Year:2015 | Volume:3 | Issue:3 (11)|Papers Count:6

Choosing a feature vector for maximizing the success of a classifier machine is very effective. In this paper, using a combination of different methods to calculate the core function, an unsupervised feature selection algorithm improvement has been proposed. Feature vector obtained by the proposed algorithm, will maximizes output accuracy of back propagation neural network classifier. In this paper we used case study of standard encoding of images compressed by alternate method and uncompressed images classifying based on their relative bit stream. Standards for classifications are JPEG and JPEG2000 and for uncompressed images is TIFF format. Using this feature vector obtained by the proposed algorithm, classifier accuracy will be about 98%.

Elliptic curve cryptography system due to the short key length and high level of security is most important encryption system for use in electronic voting. The problem with this system is a lot of computation time due to the complexity of computational operations on elliptic curve is over. Multiplication of elliptic curve cryptography system is time consuming operations that about 85% of the time spent implementing the encryption algorithm stems. Hence, we propose an optimal method to reduce the cost of providing time of multiplication operations. The proposed method improved in two main parts, the parts of the control and computing encryption algorithm, has the good performance. The result of evaluation and comparison of the proposed method with some conserned algorithms, shows that this method compared to other algorithms, is faster and very good performance.

Cyber-attacks against the web-based military command systems is very common in the age of electronic warfare. Web application is one of the most widely used tools in the world wide web. Because of its dynamic nature, it is vulnerable to serious security risks. Web-based command and control systems security considerations are very important for the modern military managers. Anomaly based intrusion detection is an approach that focuses on new and unknown attacks.A method for anomaly detection in web applications using a combination of one-class classifiers, is proposed. First, in preprocessing phase, normal HTTP traffic is logged and Features vector is extracted from each HTTP request. The proposed method consists of two steps; In the training phase, the extracted features vectors associated with each request, enter the system and the model of normal requests, using combination of one-class classifiers, is learned. In the detection phase, anomaly detection operation is performed on the features vector of each each HTTP request using learned model of the training phase. S-OWA operator is used to combine the one-class classifiers. The data used for training and test are from CSIC2012 dataset. Detection rate and false alarm rate obtained from experiments, shows better results than other methods.

The exact location of wireless communication systems aimed at both military and civilian applications is an important issue. In wireless communication networks, the user's location is done by using the measured signals time of arrival (TOA) from MS to BS. One of the most accurate methods for determining the TOA -based location systems is the use of neural networks.In this paper, a new algorithm is provided to improve the accuracy of locating based on BP neural network. In the newly proposed algorithm, instead of possible crossing points of TOA circles as neural network input, radius of TOA circle is used that much less than the previous method of complexity and accuracy is increased. Analysis and simulations show reduction positioning errors to a greater extent than half in the new algorithm compared to the previous method and it shows the location carefully.

In recent years, cyber security has become one of the main objectives of military organizations. On the other hand, forces identification, authentication and their security have become one of the basic needs of military centers. Although data encryption prevents user access to data contents, an attacker can forges exchanged data by access to communications channels. As a result, providing secure protocols for authentication systems, to prevent different attacks is very important. In this paper, we cryptanalyze a mutual RFID authentication protocol (SPRS) that presented in 2013. Unlike climes of the designers of protocol, we show that their protocol has some weaknesses yet and does not secure against some attacks such as rival secret values, tag impersonation and tractability. Then, an improved version of SPRS protocol is proposed that eliminates SPRS weaknesses. Also, the security and the privacy of proposed protocol are compared with some mutual authentication protocols that proposed recently.

red memory, online search time and success probability are the main performance metrics of a Time Memory Trade-Off (TMTO) Attack. One of the basic challenges in TMTO attack is the way of choosing TMTO attack parameters like number and length of chains to meet some certain performance metrics. Considering online breaking time, we propose an optimized procedure for selecting rainbow table TMTO attack parameters. Unlike previous works that mainly deal with minimizing required memory in the rainbow table TMTO attack, we simultaneously focus on the required memory and the online breaking time and consider index and sequential search techniques. Our parameter selection technique is optimized regarding the minimization of the required memory subject to a certain success probability and a maximum online breaking time. Obtained results are two compact mathematical expressions for determining the rainbow table TMTO attack parameters, number and length of chains for the sequential and the index search methods. The application of our optimized parameter selection procedure is also shown in few sample design examples.