نشریه مدیریت مهندسی و رایانش نرم
سال هشتم شماره 1 (پیاپی 14، بهار و تابستان 1401)

Today, Most important issues in the industry of non-linear and multi-parametric are considered optimization problems. On the other hand, the attractiveness of the behavior and interaction of animals has led the computer scientists, inspired by these interactions to create algorithms for optimization problems, which in many cases provide quick and acceptable solutions to complex problems. One of the propagation intelligence algorithms is firefly algorithm, which is bace on the exposure of luminous worms and their absorption into more light. The main problem with algorithms Such as firefly is that takes a lot of time to convege the desired answers. So if the number of firefly worms is more than 128, their run time with CPU is 2.5820 milliseconds but with using GPU 1.5090 milliseconds. In this paper, we intend to use a pc graphics unit to provide a version of the firefly algorithm that converages to the desired solutions more quickly while maintaining accuracy.

For each software product essential elements in software engineering includes quality, cost and time. Reusability of components can reduce development time and costs and increase product quality. Components used and tested in the past and, can be used and recommended at present too. Evaluation of software reusability is important as evaluation of other qualitative characteristics, and set of methods and criteria are provided for it. In this paper, the optimal set of evaluation criteria for more effective evaluation of reusability of web-based applications has been extracted and recognized. To achieve the optimal set of criteria, the previous criteria has been checked. Then optimum set of criteria has been selected. In other words, based on the criteria for optimal selection criteria, minimum set of criteria that can be fully cover all aspects of web-based software reusability have been recognized and proposed. The performance evaluation of proposed optimal set of selected criteria with content management system, reusability evaluation for every software module could well be done by the developers and the testing system.

Diabetes is a disease which, as well as prevention, requires a high level of care, such as monitoring the blood sugar changes. The timely diagnosis of disease plays an important role in its treatment and decreases the damage caused by the disease. Therefore, it is essential to diagnose diabetes. Since hybrid algorithms have a high ability to predict and diagnose various diseases, this article presents an intelligent approach to the diagnosis of this disease, using a hybrid algorithm of flower pollination and K-nearest neighbor ensemble. The accuracy of the proposed method is measured to be 97.78, by using Pima Indians Diabetes (PID) dataset, consisting of 768 samples and 8 features. The results show that the accuracy of this approach has significantly increased compared with the previous studies, and confirms the superiority of the proposed method.

This study analyzes the portfolio optimization model, by considering the financial management and investment science in order to evaluate risks and return in regard with restrictions such as buyers’ assets for purchasing per share. Accordingly, a novel model is designed as linear programming in order to optimize the investment portfolio, considering the expected rate of return, the minimum risk, and the buyer’s assets. After the introduction of the model as linear programming and expressing the related limitations, different types of investments which an investor can consider in order to form an investment portfolio were studied. Finally, an approach is proposed to solve the model by using the genetic algorithm, and is implemented and analyzed in regard with a real example. According to the results of this study, the new model reduced downside risk in comparison with previously proposed models, in a manner that its stair descent continues as the number of shares under study increases.

In this research, a novel method is proposed to optimize the costs associated with the supply chain of short-lived blood products, which is different from other existing methods. According to the novelty of this approach and the lack of familiarity with its difficulties, a scaling is applied in order to reduce the size of problem space, which can lower the accuracy of the solution. On the other hand, with regard to the problem solving dimensions, the number of solution iterations would be limited. Ultimately, after solving the above problems, the collected data are entered into the formulation of Markov’s decision-making process and solved using the successive approximation approach. The solution of this approach helps the decision-maker to choose one of (LIFO-LIFO), (FIFO-FIFO), and (LIFO, FIFO) policies. By investigating different costs and comparing various (LIFO-LIFO), (FIFO-FIFO), and (LIFO, FIFO) policies, it can be concluded over iterations that policies (LIFO-LIFO) and (LIFO -FIFO) policies have better performances than the (FIFO-FIFO) policy, and will cost much less in the long run.

Today, protecting data privacy is considered as the most important challenge in the Internet of Things networks. These networks contain important data which is transmitted among the nodes. For this reason, it is very important to pay attention to the concern of data privacy in the Internet of Things. Studies show that the lack of a complete, efficient and standard model has caused many security flaws in these networks. By using the Internet limitations, this study discusses how to create a target function to protect data privacy in the Internet of Things. The proposed approach aims to find an optimal path with high competence. This method has been implemented in MATLAB software and the results of experiments show that although the Artificial Immune System (AIS) algorithm has no certain solution for optimization problems, it finds a near optimal answer. In addition, the response time of AIS algorithm is lower compared with the examined algorithms. According to the convergence diagrams, the AIS algorithm has good convergence.

Recently, there has been a remarkable growth of research on the practical applications of game theory in networks, and in particular, the modeling of users’ behavior in distributed and decentralized systems. Reducing the runtime of operations in these types of systems will increase their performance. In order to achieve this goal, the system can be implemented using an object-oriented approach, through which the client machine treats the srver machine as an object, and the communication between them is done only through a proxy. In these types of systems, users have a set of possible choices, and may choose personal benefits over the interest of the whole system and other users. Since in a distributed system, all users want to control their resource of choice, the use of game theory can be a good tool to evaluate the behavior of selfish nodes. In this paper, game theory is used to investigate the behavior of nodes in an object-oriented distributed system, in which the communication between the client machine and the server machine is established through a proxy. To understand the behavior of nodes in a distributed system,one-time games and infinitely-repeated games are studied, and finally, the behavior of one node against an object-oriented distribution system is analyzed. According to the results of this study, nodes defect and will be uncooperative in one-time games. But when there is a strategy of an infinitely-repeated game, the cooperation between nodes will depend on the discount factor, or the probability of the next stage.

The proposed control scheme in this paper is based on the design of an optimal load distribution of reactive power for High-Voltage Direct-Current systems (HVDCs). The purpose is to monitor and implement appropriate operations to create a balance in the power and eliminate the voltage or current that has exceeded the permissible limit. Two-terminal HVDC transmission lines are of the most important elements of the power system. The optimization of reactive power distribution for power systems is defined based on the minimization of power loss under equal and unequal constraints. The problem of optimizing the reactive power distribution in integrated AC-DC power systems in the form of HVDC lines is generalized with regard to the control specifications for power transmission. In this paper, the problem is solved using Genetic Optimization Algorithm (GA). The validation, efficiency and the effectiveness of the proposed method are measured by comparing the results of the GA and those obtained by previous researches. Results indicate the effectiveness of the proposed method in the optimization of power distribution and the design of HVDC lines.

Through virtualization technology, current cloud data centers are becoming more flexible and secure, and are allocated on demand. A key technology playing an important role in cloud data centers is the resource scheduling program. In this paper, a near-optimal strategy is proposed to solve the problems in this field, by using an evolutionary particle swarm algorithm to reduce the range of multiple targets to a proper level. The placement method based on the particle swarm optimization algorithm can act as real-time placement, due to the increase in computational capability of processors over the past five years. This placement is a searching method in which competencies are dynamically altered based on the variance of fitness values ​​in each generation. This migration and placement approach also minimizes the completion time for virtual machines. In order to assess the proposed method, the results were analyzed and compared through various qualitative criteria, from different aspects and based on changes in different functioning parameters. The performance of the proposed method was compared with other approaches in this field and reflects the high quality of the proposed method.

An analysis of social networks is necessary to detect anomalous users, due to the popularity of these networks. This paper aims to detect anomalous users in location based social networks. For this purpose, an ego graph is computed for each user, and the five variables vertex degree, edge size, edge weight, betweenness centrality and eigenvector centrality are calculated with respect to the weights of the edges in this graph. Then six relationships between two of these variables are made up, and for each of these relationships, the line equation is obtained in the coordinate system of the two variables. This equation is used to predict the value of the variables. Based on this predicted value, the user's score is determined, and anomalous users are detected. The proposed method investigates anomalies in the friendship graph, location of residence and interests of users. The results indicate that the proposed method has been able to detect anomalous users by examining the scores of star and clique structures in the graph.

With regard to the entry of most banks into the competitive environment of universal banking, as a customer-centric approach aiming to provide financial services to customer groups, it is important to assess the performance of the business process in bank branches using this approach. Due to the uncertainty and desirability of some indicators as well as the existence of shared resources in the real environment of modelling, this paper aims to design a network fuzzy data envelopment analysis (DAE) model, with the simultaneous consideration of undesirable variables and shared resources, and its practical use to measure the effectiveness of service delivery process in universal banking. The results indicate how the branches performed in each of the processes of universal banking. The lower mean efficiency of service provision to customers reflects the necessity for the bank to provide a portfolio of products and better credit services. Comparing the performance of branches in terms of each process can play an important role in the adoption of incentive and punitive policies for managers.