Computing最新文献

One of the most popular models that provide computer resources today is cloud computing. Today’s dynamic and successful platforms are created to take advantage of various resources available from service providers. Ensuring the performance and availability of such resources and services is a crucial problem. Any software system may be subject to faults that might propagate to cause failures. Such faults with the potential of contributing to failures are critical because they impair performance and result in a delayed reaction, which is regarded as a dependability problem. To ensure that critical faults can be discovered as soon as possible, the impact of such faults on the system must be tested. The performance and dependability of cloud-native systems are examined in this empirical study using fault injection, one of the chaos engineering techniques. The study explores the impacts and results of injecting various delay times into two cloud-native applications with diverse user numbers. The performance of the applications with various numbers of users is measured in relation to these delays, which accordingly reflects measuring the dependability of those systems. Firstly, the systems’ architecture were identified, and serverless with two Lambda functions and containerised microservices applications were chosen, which depend on utilising and incorporating cloud-native services. Secondly, faults are injected in order to quantify performance attributes such as throughput and latency. The results of several controlled experiments carried out in real-world cloud environments provide exploratory empirical data, which promoted comparisons and statistical analysis that we utilised to identify the behaviour of the application while experiencing stress. Typical results from this investigation include an overall reduction in performance that is embodied in an increase in latency with injecting delays. However, a remarkable result is noticed at a particular delay in which defects and availability problems appear out of nowhere. These findings assist in highlighting the value of using chaos engineering in general and fault injection in particular to assess the dependability of cloud-native applications and to find unpredicted failures that could arise quickly from defects that aren’t supposed to spread and result in dependability issues.

Swarm intelligence (SI) algorithms represent a class of Artificial Intelligence (AI) optimization metaheuristics used for solving complex optimization problems. However, a key challenge in solving complex problems is maintaining the balance between exploration and exploitation to find the optimal global solution and avoid local minima. This paper proposes an innovative Swarm Intelligence (SI) algorithm called the Four Vector Intelligent Metaheuristic (FVIM) to address the aforementioned problem. FVIM’s search strategy is guided by four top-performing leaders within a swarm, ensuring a balanced exploration-exploitation trade-off in the search space, avoiding local minima, and mitigating low convergence issues. The efficacy of FVIM is evaluated through extensive experiments conducted over two datasets, incorporating both qualitative and quantitative statistical measurements. One dataset contains twenty-three well-known single-objective optimization functions, such as fixed-dimensional and multi-modal functions, while the other dataset comprises the CEC2017 functions. Additionally, the Wilcoxon test was computed to validate the result’s significance. The results illustrate FVIM’s effectiveness in addressing diverse optimization challenges. Moreover, FVIM has been successfully applied to tackle engineering design problems, such as weld beam and truss engineering design.

The optimal foraging algorithm (OFA) is a swarm-based algorithm motivated by animal behavioral ecology theory. When solving complex optimization problems characterized by multiple peaks, OFA is easy to get trapped in local minima and encounters slow convergence. Therefore, this paper presents an improved optimal foraging algorithm with social behavior based on quasi-opposition (QOS-OFA) to address these problems. First, quasi-opposition-based learning (QOBL) is introduced to improve the overall quality of the population in the initialization phase. Second, an efficient cosine-based scale factor is designed to accelerate the exploration of the search space. Third, a new search strategy with social behavior is designed to enhance local exploitation. The cosine-based scale factor is used as a regulator to achieve a balance between global exploration and local exploitation. The proposed QOS-OFA is compared with seven meta-heuristic algorithms on a CEC benchmark test suite and three real-world optimization problems. The experimental results show that QOS-OFA is better than other competitors on most of the test problems.

Service composition is widely employed in cloud manufacturing. Due to the abundance of similar cloud manufacturing services, the search space for optimizing service composition tends to be expansive. Existing optimization models primarily focus on QoS (quality of service) while often neglecting QoC (quality of collaboration). Furthermore, there remains scope for improving the quality and stability of service composition optimization. Therefore, this paper proposes a two-phase method for optimizing service composition in cloud manufacturing. In the first phase, we introduce a service cluster-oriented service response framework, efficiently generating the candidate response service set to reduce solution search space. In the second phase, we construct an optimization model that integrates QoS and QoC. Subsequently, we devise an artificial bee colony (ABC) algorithm incorporating a multi-search strategy island model to optimize cloud manufacturing service composition. Experimental results demonstrate that the introduction of service clusters enhances search efficiency, with the proposed method outperforming compared ABC algorithms and other swarm intelligence algorithms in optimization quality and stability.

The identification of vital spreaders in complex networks has been one of the most interesting topics in network science. Several methods were proposed to deal with this challenge, but there still exist deficiencies in previous methods, such as excessive time complexity, inadequate accuracy of recognition results after dividing the topological structure, and the ignorance of neighbors’ attribute information in the links’ significance model. To address these issues and promote identifying ability more effectively, the proposed extended centrality upon hybrid information, named EISMC, introduces the interpretative structure model (ISM) and improves hierarchical weight results after the division in hierarchies. Based on the hierarchical structure of Improved Kshell decomposition (IKs), the weight value of each layer is updated, and meanwhile the local centrality under link significance (LinkC) is created to supplement local features in this method. In this paper, six real-world networks and nine comparison methods are applied to conduct a series of simulations and tests. Results demonstrate that the proposed method outperforms state-of-the-art algorithms in the identifying effects for good spreading influence.

Indoor positioning has been extensively studied for at least the past twenty years. In the list of the most common solutions, those based on the Received Strength Signal Indicator (RSSI) have gained importance due to the simplicity of RSSI as well as the fact that it is available in several wireless sensor networks. In this work, we propose SeALS (Selection Strategy of Access Points with Least Squares Estimation), a new RSSI-based indoor positioning system using Bluetooth Low-Energy (BLE) access points, whose accuracy is improved by a new selection strategy of collected RSSI combined with the Ordinary Least Squares (OLS) estimation method. The main advantage of the proposed solution is the fact that it requires less time in the training phase allied with better system accuracy if compared to traditional methods. The proposed system is validated in a large-scale, real-world scenario, and the obtained results for the positioning error are reduced by up to 13% concerning the pure OLS method, and by up to 30% concerning the widely deployed K-Nearest Neighbors technique.

The proliferation of available Web services presents a big challenge in selecting suitable services. Various methods have been devised to predict Quality of Service (QoS) values, aiming to address the service selection problem. However, these methods encounter numerous limitations that hinder their prediction accuracy. A key issue stems from the dynamic nature of the service environment, leading to fluctuations in QoS values due to factors like network load and hardware issues. To mitigate these challenges, QoS selection methods have leveraged contextual information from the surrounding environments, such as service invocation time, user, and service locations. Among these methods, Collaborative Filtering (CF) has gained notable importance. In recent years, several CF methods have incorporated service invocation time into their prediction processes, giving rise to what is commonly known as time-aware CF methods. Despite the increasing adoption of time-aware CF methods, there remains a notable absence of a dedicated and comprehensive literature review on this topic. Addressing this gap, this paper conducts an analysis of the literature, reviewing the forty (40) most prominent studies in this domain. It offers a thematic categorization of these studies along with an insightful analysis outlining their objectives, advantages, and limitations. The review also identifies key research gaps and proposes potential directions for future investigations. Overall, this literature review serves as an up-to-date resource for researchers engaged in service-oriented computing research.

The motivation of task scheduling in heterogeneous computing systems is the optimal management of heterogeneous distributed resources as well as the exploitation of system capabilities. Energy consumption is one of the most important issues in dealing with task scheduling in heterogeneous distributed systems. In addition to energy, the task completion time and the task cost have also been added to the concerns of the users. Since the nature of computing systems is heterogeneous and dynamic, task scheduling with traditional methods is inefficient. Meta-heuristic approaches for task scheduling in heterogeneous distributed systems are open problems that have attracted the attention of researchers. So far, many meta-heuristic approaches have addressed the task scheduling problem. However, most of these algorithms are developed for homogeneous systems and optimize only one of the quality-of-service parameters. With this motivation, this paper presents an optimization for energy-aware design of task scheduling in heterogeneous distributed systems using meta-heuristic approaches. We simultaneously consider several parameters such as energy, task completion time and task execution cost for task scheduling. The Harris Hawk Optimization (HHO) algorithm is considered for the optimization task due to its adaptability to large search spaces. We combine HHO with a greedy algorithm to avoid local optima and early convergence. The evaluation of the proposed method has been done through numerical simulations. Experimental results show promising performance of the proposed method in terms of energy consumption.

With the widespread use of attribute networks, anomalous node detection on attribute networks has received increasing attention. By utilizing communities as reference contexts for local anomaly node detection, it is possible to uncover a multitude of significant anomalous nodes. However, most of the current methods that use communities as reference context of anomalous nodes usually do not consider the accuracy of the reference context. The rough classification results obtained from community detection are used as reference contexts for anomalous node detection. The possibility of errors occurring in the reference context may subsequently result in detection errors for anomalous nodes. Based on this, we propose an integrated framework named ADRC (Anomaly Detection in attribute networks based on Refining Context) to simultaneously perform anomalous node detection and detailed adjustment of reference contexts. Meanwhile, to better reflect the anomaly degree of the nodes, we design an evaluation metric and rank the anomalous nodes by it. Comparisons are made with state-of-the-art algorithms on publicly available datasets and the results show that our approach has significant advantages.

Virtual Machine Placement (VMP) plays a significant role in improving efficiency of Cloud Data Center (CDC). With the dramatic increase in the use of cloud computing, it seems necessary to apply effective algorithms to reduce the power consumption of CDC. VMP is known as a NP-Hard problem that cannot be solved by deterministic algorithms in polynomial time. In this paper, an algorithm named Combinated Random Best First Fit (CRBFF) is proposed with the aim of increasing the Quality of Service (QoS), in which Virtual Machines (VMs) are optimally placed on heterogeneous Physical Machines (PMs). The effectiveness of CRBFF is evaluated by different metrics on Google Compute Engine (GCE), Amazon Web Service Elastic Compute Cloud (AWS EC2) and Microsoft Azure scenarios and the results show that CRBFF performs better than other common algorithms.