JOURNAL OF ADVANCES IN COMPUTER RESEARCH
Year:2015 | Volume:6 | Issue:4 (22)|Papers Count:10

Congestive heart failure (CHF) remains to be one of the major cardiovascular disorders in the world. Due to the prevalence of CHF related issues, it is prudent to seek out new prognostic predictors that would facilitate the prevention, monitoring, and treatment of the disease on a daily basis. A detection approach using entropy measures extracted from surface electrocardiograms (ECGs) and classification for congestive heart failure (CHF) is presented in this paper. Four different entropies are used: approximate entropy (ApEn), sample entropy (SampEn), permutation entropy (PE), and energy entropy (EE). These entropies are employed to evaluate the irregularity and complexity of ECG time series and discuss the viability of recognizing CHF patients from normal subjects. Student’s t-tests and receiver operating characteristic (ROC) plots show that among the four entropies, EE outperforms other three entropies. These tests also indicate the feasibility of using surface ECGs to effectively discriminate CHF patients from normal subjects.

Dynamic economic load dispatch is one of the most important roles of power generation’s operation and control. It determines the optimal controls of production of generator units with predicted load demand over a certain period of time. Economic dispatch at minimum production cost is one of the most important subjects in the power network’s operation, which is a complicated nonlinear constrained optimization problem. Since dynamic economic load dispatch was introduced, several intelligent methods have been used to solve this problem. In this paper, an Improved Particle Swarm Optimizer (IPSO) and Water Cycle optimizer (WCO), as swarm-based optimization algorithms, have been proposed to solve dynamic economic load dispatch problem and their results compare with each other. These algorithms are applied to a dynamic economic dispatch problem for 6-unit power systems with a 24-h load demand at each one hour time intervals. The goal of the research is categorized in two parts; first of all, introduction of application of new heuristic method for solving economic load dispatch problem and second, comparison between two swarm-based algorithms. Obtained results show that WCO is very fast and also reach to better results and minimum.

Car license plate recognition is addressed in this paper. Given the development of intelligent transportation systems, it is absolutely essential to implement a strong license plate recognition system. Efforts were made to put forward a novel reliable method for car license plate recognition in Iran. Each license plate recognition system comprises three main parts. The first part is the license plate detection stage. The blue color feature of the license plate margin along with Scale-Invariant Feature Transform (SIFT) algorithm were used for this purpose. The accuracy of the presented method over the database was approximately 90% in less than a second. License plate morphological features were utilized upon character segmentation. Using these features, areas with sizes close to that of the characters of a license plate may be searched. The accuracy of this method was almost 95%. A probabilistic neural network together with a Support Vector Machine (SVM) was employed at the character recognition stage. For this stage, an accuracy of nearly 97% in 55 milliseconds for each license plate was achieved.

Nowadays, with the advent of new satellite services, the need for resource management in the emerging fifth generation satellite systems (5G-satellite) is inevitable. Thus, to solve this problem, the Bandwidth Manager for resource reservation in satellite link is mandatory. On the other hand, due to limited resources, their resource management¬ is essential. In order to resource management in 5G-satellite systems, can be applied in one phase or many phases. In this paper, resource management for 5G-satellite services is evaluated. The proposed optimization problem is to maximize mean response time under the propagation delay constraints in satellite links. We solve the considered optimization problem via the single phase and two phase algorithms. Finally, through simulation, the proposed algorithms are investigated and confirmed. In our scenarios, satellite is a Central node in call flows and ground stations are End nodes in 5G-satellite based on Internet protocol. So we simulated all of scenarios in matlab software for this reason.

Sensor networks generally consist of a very great number of sensor nodes which will be spread into a vast environment and aggregate data out of it. The sensor nodes are afflicted with some limitations as follows memory, reception, communication as well as calculation capability, and battery power. The transmission of a great amount of extra data increases data transmission and proportionally increases the amount of energy and bandwidth for the data transmission. One solution for this issue is data aggregation. The results of aggregated data influence the accuracy and precision of the final result already gleaned from the base station. The main challenge in such networks is how to further elongate the network lifetime and among the factors doing so is the energy consumption or energy optimization. The clustering is one apt method in place for furthering the network life span. Respectively the clustering protocols have come up with a suitable method for the so called challenge or more simply put increasing the lifetime. In this paper the researchers attempt to bring forth yet another efficient protocol for data aggregation hinging around clustering which uses maximum residual energy and minimum distance for selecting the cluster-head to reduce the consumption of energy. The experimental results point to this very fact that Energy-Efficient Clustering Algorithm through Residual Energy and Average Distance (EECA-READ) attains very good performance.

Due to limitations in computing resources of mobile nodes, Mobile Ad-hoc Networks (MANETs) are very fault-prone. So far, redundancy at different levels of the network has been considered an efficient strategy to enhance the fault-tolerance of networks. Indeed, ad-hoc networks are extremely redundant; therefore, natural redundancy to fault tolerance enhancement has a great impact on the network performance. In this paper, a fault tolerant algorithm for MANETs has been proposed that by assigning backup node(s) to each node tries to increase in fault-tolerance. For this reason, the proposed algorithm chooses the backup nodes from among those nodes having the same movement route. The nodes movement route can be determined or predicted through the backup nodes table. Choosing backup nodes is done based on the time of nodes adjacency. Experimental results taken from NS-2 simulator indicated that in comparison to other algorithms, the proposed algorithm increases by 1) the package delivery rate in relation to the percentage of fault, and 2) the package delivery rate in relation to various mobile nodes’ pause time.

In this paper, a new robust controller based on geometric homogeneity and adaptive integral sliding mode is proposed for a class of second order systems. The upper bound of the system disturbances is not required. Fully unknown parameters have been considered in the described model and its finite–time convergence to zero equilibrium point is proved. Moreover, the controller is developed in the presence of control singularity and unknown non-symmetric input saturation. The finite time stability of the proposed controller has been proved via classical Lyapunov criteria. In order to tune the control parameters, all the positive constant gains are optimized by ant colony optimization algorithm during the offline input-output training data. Two polar robots are introduced to show the performance of the designed controller. The robustness and error accuracy are proved in simulation results. Moreover, the effects of input nonlinearity such as input saturation have been considered in the simulation.

 Classification Ensemble, which uses the weighed polling of outputs, is the art of combining a set of basic classifiers for generating high-performance, robust and more stable results. This study aims to improve the results of identifying the Persian handwritten letters using Error Correcting Output Coding (ECOC) ensemble method. Furthermore, the feature selection is used to reduce the costs of errors in our proposed method. ECOC is a method for decomposing a multi-way classification problem into many binary classification tasks; and then combining the results of the subtasks into a hypothesized solution to the original problem. Firstly, the image features are extracted by Principal Components Analysis (PCA). After that, ECOC is used for identification the Persian handwritten letters which it uses Support Vector Machine (SVM) as the base classifier. The empirical results of applying this ensemble method using 10 real-world data sets of Persian handwritten letters indicate that this method has better results in identifying the Persian handwritten letters than other ensemble methods and also single classifications. Moreover, by testing a number of different features, this paper found that we can reduce the additional cost in feature selection stage by using this method.

Today a significant part of available data is saved in text database or text documents. The most important thing is to organize these documents. One way to organize text documents is to classify them. To classify texts is to assign text documents to their actual categories. This has two main steps, i.e. feature- and learning algorithm selection. There have been several methods suggested to classify text documents. In this paper, we propose a combined method to do this more efficiently. When selecting features, the proposed method uses filtering in order to reduce complexity and it is implemented using naïve Bayes and decision tree categories. Results indicate advantages of this combined method to individual classifying.

Threat assessment in the computer networks of organizations can reduce damage caused by attacks and unexpected events. Data fusion models such as the JDL model provide efficient and adequate sensors to gather the right information at the right time from the right components. This information then is refined and normalized to provide situational awareness and assess events that may be intended as a threat. This study suggests a new method based on the JDL model where data collected from different sources is normalized into an appropriate format. After normalization, Data is converted into the information. Threat assessment unit analyzes this information based on various algorithms. We use three algorithms to detect anomaly, one to correlate alerts, and one to determine the successfulness of an attack. The model is then evaluated based on a small simulated network threat to ascertain the efficacy of the proposed method. The results show that the method is an appropriate model for situational awareness and threat assessment.