نشریه مهندسی برق و مهندسی کامپیوتر ایران
سال هجدهم شماره 4 (پیاپی 59، زمستان 1399)

Lack of an efficient video frame delivery method due to high delay in Pull method and large number of duplicated frames in Push method, as the two main content delivery methods among peers, has been a strong motivator for introducing hybrid methods based on these two basic approaches for live video streaming in mesh-based peer-to-peer networks. Recent studies show that these hybrid methods suffer from inherent challenges of the two basic approaches because they are just a sequential or parallel execution of them. In this regard, this research introduces AMIN, a novel hybrid method for intelligently exchanging video frames among peers. Using AMIN, contrary to Pull, each peer sends its buffer map status (BMS) to its two-hop neighbors and the peer who receives the BMS will immediately check which video frames it can send to that peer instead of requesting missed video frames in its buffer from it. In addition, contrary to Push and because of BMS, peers do not blindly send video frames to their neighbors. Simulation results show that video quality considerably increases in peers, while End-to-End delay, received delay and the number of duplicated frames decrease in comparison with two basic methods as well as another recent similar approach.

One of the applications of sensor networks is to track moving target. In designing the algorithm for target tracking two issues are of importance: reduction of energy consumption and improvement of the tracking quality. One of the solutions for reduction of energy consumption is to form a tracking cluster. Two major challenges in formation of the tracking cluster are when and how it should be formed. To decrease the number of messages which are exchanged to form the tracking cluster an auction mechanism is adopted. The sensor’s bid in an auction is dynamically and independently determined with the aim of establishing an appropriate tradeoff between network lifetime and the accuracy of tracking. Furthermore, since the tracking cluster should be formed and activated before the target arrives to the concerned region (especially in high speed of target), avoidance from delay in formation of the tracking cluster is another challenge. Not addressing the mentioned challenge results in increased target missing rate and consequently energy loss. To overcome this challenge, it is proposed to predict the target’s position in the next two steps by using neural network and then, simultaneously form the tracking clusters in the next one and two steps. The results obtained from simulation indicate that the proposed algorithm outperforms AASA (Auction-based Adaptive Sensor Activation).

Software-defined network (SDN) is the next generation of network architecture thatby separating the data plane and the control plane enables centralized control with the aim of improving network management and compatibility. However, due to the centralized control policy, this type of network is prone to Inaccessibility of control plane against a denial of service (DoS) attack. In the reactive mode, a significant increase in events due to the entry of new flows into the network puts a lot of pressure on the control plane. Also, the presence of recurring events such as the collection of statistical information from the network, which severely interferes with the basic functionality of the control plane, can greatly affect the efficiency of the control plane. To resist attack and prevent network paralysis, this paper introduces a new architecture called SAHAR, which consists of a control box consisting of a coordinator controller, a primary flow setup controller, and one or more (as needed) secondary flow setup controller(s). Assigning monitoring and managing tasks to the coordinator controller reduces the load of flow setup controllers. In addition, dividing the incoming traffic between the flow setup controllers by the coordinator controller distributes the load at the control plane. Thus, by assigning the traffic load resulting from a denial-of-service attack to one or more secondary flow setup controller(s), the SAHAR architecture can prevent the primary flow setup controller from impairment and resist DoS attacks. Tests show that SAHAR performs better in the face of a DoS attack than existing solutions.

Today, the rapid growth in cloud computing resources usage has increased energy consumption in data centers. Green cloud computing goal is to decrease the energy consumption of data centers. In the meantime, service aggregation is a good method to reduce energy consumption in these systems. Existing aggregation methods with unnecessary migration, the unbalanced workload of hosts, and ignoring the relationship between services may reduce the quality of service and increase energy consumption. Therefore, in this study, by migrating the necessary services based on priority (including the number of children, the level and communication cost of each service), from hosts with the unbalanced workload to hosts that contain partner services, the productivity of available resources is improved and the energy consumption is decreased. Live services migration based on prioritizing and minimizing the number of migrations can also lead to response time decrease and system efficiency increase. The proposed method can lead to an 11.79% decrease in energy consumption, a 12.15% reduction in the number of service migrations, and a 1.55% increase in the number of hosts that have been shut down.

Nowadays, one of the most important issues in the field of image processing is image de-blurring. De-blurring of an image can be achieved via two different approaches; blind de-blurring and non-blind de-blurring. In blind de-blurring, the kernel by which the blur has occurred is assumed unknown, while in non-blind de-blurring, this kernel is given. In blind de-blurring, the blurring kernel must be estimated in order to sharpen the corrupted image. This may increase the computational cost of the de-blurring process. Non-blind image de-blurring is an ill-posed problem with linear reverse issues. Therefore, we develop optimization problems in order to estimate the original sharp images. Usually, non-blind de-blurring methods assume that the blurring kernel is error-free, however, in practice our knowledge of the PSF is uncertain. Hence, in this paper, we use a semi-blind method for de-blurring the blurred image that is robust to this uncertainty. The proposed robust optimization model is followed by a filter for image de-blurring that can attain the solution with lowest possible error in the worst case scenarios, that is, the maximum uncertainty about the blurring kernel. Based on the simulation results, our proposed semi-blind model yields more than 4 dB PSNR improvements compared to conventional blind image de-blurring methods.

Today, in many real-world applications, such as social networks, we are faced with data streams which new data is appeared every moment. Since the efficiency of most data mining algorithms decreases with increasing data dimensions, analysis of the data has become one of the most important issues recently. Online stream feature selection is an effective approach which aims at removing those of redundant features and keeping relevant ones, leads to reduce the size of the data and improve the accuracy of the online data mining methods. There are several critical issues for online stream feature selection methods including: unavailability of the entire feature set before starting the algorithm, scalability, stability, classification accuracy, and size of selected feature set. So far, existing methods have only been able to address a few numbers of these issues simultaneously. To this end, in this paper, we present an online feature selection method called MMIOSFS that provides a better tradeoff between these challenges using Mutual Information. In the proposed method, first the feature set is mapped to a new feature using joint Random variables technique, then the mutual information of new feature with the class label is computed as the degree of relationship between the features set. The efficiency of the proposed method was compared to several online feature selection algorithms based on different categories. The results show that the proposed method usually achieves better tradeoff between the mentioned challenges.

Multi-cloud environments consist of the considerable variety of resources where the cost of scheduling workflow applications can be significantly reduced in such environments and the resource limitationsimposed by commercial cloud providers can bealso overcome. Accordingly, this study addresses the scheduling of scientific workflowapplications in a multi-cloud environment under a deadline with the aim of minimizing costs. In this paper,an algorithm for scheduling of workflow applications in multi-cloud environment is presented using the cuckoo search algorithm which is one of the most popular meta-heuristic methods. The Cuckoo Search Algorithm is able to search the solution space in a short time and find solutions in the vicinity of the optimal global solution that is close to it. The results show that the proposed approach of this research has better performance in comparison with other meta- heuristic approach in terms of cost reduction. Moreover, the obtained solutions of the proposed meta- heuristic algorithm are in a desirable degree close to the global optimal solutions of mathematical model.