Journal of Network and Computer Applications最新文献

Cloud gaming (CG), as an emergent computing paradigm, is revolutionizing the gaming industry. Currently, cloud gaming service providers (CGSPs) begin to integrate edge computing with cloud to provide services, with the aim of maximizing gaming service revenue while considering the costs incurred and the benefits generated. However, it is non-trivial to maximize gaming service revenue, as future requests are not known beforehand, and poor resource provisioning may result in exorbitant costs. In addition, the edge resource provisioning (ERP) problem in CG necessitates a trade-off between cost and inevitable queuing issues in CG systems. To address this issue, we propose ERPOL (ERP Online), a convenient and efficient approach to formulate ERP strategies for CGSPs, without requiring any future information. The performance of ERPOL has been theoretically validated and experimentally evaluated. Experiments driven by real-world traces show that it can achieve significant cost savings. The proposed approach has the potential to transform how CGSPs manage their infrastructure.

To accelerate the convergence speed and improve the accuracy of the federated shared model, this paper proposes a Federated Transfer Learning method based on frozen network parameters. The article sets up frozen two, three, and four layers network parameters, 8 sets of experimental tasks, and two target users for comparative experiments on frozen network parameters, and uses homomorphic encryption based Federated Transfer Learning to achieve secret transfer of parameters, and the accuracy, convergence speed, and loss function values of the experiment were compared and analyzed. The experiment proved that the frozen three-layer network parameter model has the highest accuracy, with the average values of the two target users being 0.9165 and 0.9164; The convergence speed is also the most ideal, with fast convergence completed after 25 iterations. The training time for the two users is also the shortest, with 1732.0s and 1787.3s, respectively; The loss function value shows that the lowest value for User-II is 0.181, while User-III is 0.2061. Finally, the unlabeled and non-empty enterprise credit data is predicted, with 61.08% of users being low-risk users. This article achieves rapid convergence of the target network model by freezing source domain network parameters in a shared network, saving computational resources.

In the Internet of Things (IoT) era, wireless sensor networks play a critical role in communication systems. One of the most crucial problems in wireless sensor networks is the sensor deployment problem, which attempts to provide a strategy to place the sensors within the surveillance area so that two fundamental criteria of wireless sensor networks, coverage and connectivity, are guaranteed. In this paper, we look to solve the multi-objective deployment problem so that area coverage is maximized and the number of nodes used is minimized. Since Harmony Search is a simple yet suitable algorithm for our work, we propose Harmony Search algorithm along with various enhancement proposals, including heuristic initialization, random sampling of sensor types, weighted fitness evaluation, and using different components in the fitness function, to provide a solution to the problem of sensor deployment in a heterogeneous wireless sensor network where sensors have different sensing ranges. On top of that, the probabilistic sensing model is used to reflect how the sensors work realistically. We also provide the extension of our solution to 3D areas and propose a realistic 3D dataset to evaluate it. The simulation results show that the proposed algorithms solve the area coverage problem more efficiently than previous algorithms. Our best proposal demonstrates significant improvements in coverage ratio by 10.20% and cost saving by 27.65% compared to the best baseline in a large-scale evaluation.

In the rapidly changing cybersecurity landscape, threat hunting has become a critical proactive defense against sophisticated cyber threats. While traditional security measures are essential, their reactive nature often falls short in countering malicious actors’ increasingly advanced tactics. This paper explores the crucial role of threat hunting, a systematic, analyst-driven process aimed at uncovering hidden threats lurking within an organization’s digital infrastructure before they escalate into major incidents. Despite its importance, the cybersecurity community grapples with several challenges, including the lack of standardized methodologies, the need for specialized expertise, and the integration of cutting-edge technologies like artificial intelligence (AI) for predictive threat identification. To tackle these challenges, this survey paper offers a comprehensive overview of current threat hunting practices, emphasizing the integration of AI-driven models for proactive threat prediction. Our research explores critical questions regarding the effectiveness of various threat hunting processes and the incorporation of advanced techniques such as augmented methodologies and machine learning. Our approach involves a systematic review of existing practices, including frameworks from industry leaders like IBM and CrowdStrike. We also explore resources for intelligence ontologies and automation tools. The background section clarifies the distinction between threat hunting and anomaly detection, emphasizing systematic processes crucial for effective threat hunting. We formulate hypotheses based on hidden states and observations, examine the interplay between anomaly detection and threat hunting, and introduce iterative detection methodologies and playbooks for enhanced threat detection. Our review encompasses supervised and unsupervised machine learning approaches, reasoning techniques, graph-based and rule-based methods, as well as other innovative strategies. We identify key challenges in the field, including the scarcity of labeled data, imbalanced datasets, the need for integrating multiple data sources, the rapid evolution of adversarial techniques, and the limited availability of human expertise and data intelligence. The discussion highlights the transformative impact of artificial intelligence on both threat hunting and cybercrime, reinforcing the importance of robust hypothesis development. This paper contributes a detailed analysis of the current state and future directions of threat hunting, offering actionable insights for researchers and practitioners to enhance threat detection and mitigation strategies in the ever-evolving cybersecurity landscape.

In this paper, we address the challenge of minimizing system energy consumption for task offloading within non-line-of-sight (NLoS) mobile edge computing (MEC) environments. Our approach integrates an active reconfigurable intelligent surface (RIS) and employs a hybrid transmission scheme combining time division multiple access (TDMA) and non-orthogonal multiple access (NOMA) to enhance the quality of service (QoS) for user task offloading. The formulation of this problem as a non-convex optimization issue presents significant challenges due to its inherent complexity. To overcome this, we introduce an innovative method termed element refinement-based differential evolution (ERBDE). Initially, through rigorous theoretical analysis, we optimally determine the allocation of local computation resources, computation resources at the base station (BS), and transmit power of users, while maintaining fixed values for the offloading ratio, amplification factor, phase of the reflecting element, and the transmission period. Subsequently, we employ the differential evolution (DE) algorithm to iteratively fine-tune the offloading ratio, amplification factor, phase of the reflecting element, and transmission period towards near-optimal configurations. Our simulation results demonstrate that the implementation of active RIS-supported task offloading utilizing the hybrid TDMA-NOMA scheme results in an average system energy consumption reduction of 80.3%.

With the rapid growth of data volume, the escalating complexity of data businesses, and the increasing reliance on the Internet for daily life and production, the scale of data centers is constantly expanding. The data center network (DCN) is a bridge connecting large-scale servers in data centers for large-scale distributed computing. How to build a DCN structure that is flexible and cost-effective, while maintaining its topological properties unchanged during network expansion has become a challenging issue. In this paper, we propose an expandable and cost-effective DCN, namely HHCube, which is based on the half hypercube structure. Further, we analyze some characteristics of HHCube, including connectivity, diameter, and bisection bandwidth of the HHCube. We also design an efficient algorithm to find the shortest path between any two distinct nodes and present a fault-tolerant routing scheme to obtain a fault-tolerant path between any two distinct fault-free nodes in HHCube. Meanwhile, we present two local diagnosis algorithms to determine the status of nodes in HHCube under the PMC model and MM* model, respectively. Our results demonstrate that despite the presence of up to 25% faulty nodes in HHCube, both algorithms achieve a correct diagnosis rate exceeding 90%. Finally, we compare HHCube with state-of-the-art DCNs including Fat-Tree, DCell, BCube, Ficonn, and HSDC, and the experimental results indicate that the HHCube is an excellent candidate for constructing expandable and cost-effective DCNs.

The Consensus algorithm is the core of the permissioned blockchain, it directly affects the performance and scalability of the system. Performance is limited by the computing power and network bandwidth of a single leader node. Most existing blockchain systems adopt mesh or star topology. Blockchain performance decreases rapidly as the number of nodes increases. To solve this problem, we first design the n-k cluster tree and corresponding generation algorithm, which supports rapid reconfiguration of nodes. Then we propose the Zebra consensus algorithm, which is a cluster tree-based consensus algorithm. Compared to the PBFT, it has higher throughput and supports more nodes under the same hardware conditions. However, the tree network topology enhances scalability while also increasing latency among nodes. To reduce transaction latency, we designed the Pipeline-Zebra consensus algorithm that further improves the performance of blockchain systems in a tree network topology through parallel message propagation and block validation. The message complexity of the algorithm is O(n). Experimental results show that the performance of the algorithm proposed in this paper can reach 2.25 times that of the PBFT algorithm, and it supports four times the number of nodes under the same hardware.

This study introduces a groundbreaking method for predicting network quality in LTE and 5G environments using only GPS data, focusing on pinpointing specific locations within a designated area to determine network quality as either good or poor. By leveraging machine learning algorithms, we have successfully demonstrated that geographical location can be a key indicator of network performance. Our research involved initially classifying network quality using traditional signal strength metrics and then shifting to rely exclusively on GPS coordinates for prediction. Employing a variety of classifiers, including Decision Tree, Random Forest, Gradient Boosting and K-Nearest Neighbors, we uncovered notable correlations between location data and network quality. This methodology provides network operators with a cost-effective and efficient tool for identifying and addressing network quality issues based on geographic insights. Additionally, we explored the potential implications of our study in various use cases, including healthcare, education, and urban industrialization, highlighting its versatility across different sectors. Our findings pave the way for innovative network management strategies, especially critical in the contexts of both LTE and the rapidly evolving landscape of 5G technology.

The proliferation of IoT devices has led to a surge in network traffic, resulting in higher energy usage and response delays. In-network caching has emerged as a viable solution to address this issue. However, caching IoT data faces two key challenges: the transient nature of IoT content and the unknown spatiotemporal content popularity. Additionally, the use of a global view on dynamic IoT networks is problematic due to the high communication overhead involved. To tackle these challenges, this paper presents an adaptive management approach that jointly optimizes caching and communication in IoT networks using a novel bi-level control method called BC3. The approach employs two types of controllers: a global ILP-based optimal controller for long-term timeslots and local learning-based controllers for short-term timeslots. The long-term controller periodically establishes a global cache policy for the network and sends specific cache rules to each edge server. The local controller at each edge server solves the joint problem of bandwidth allocation and cache adaptation using deep reinforcement learning (DRL) technique. The main objective is to minimize energy consumption and system response time with utilizing the global and local observations. Experimental results demonstrate that the proposed approach increases cache hit rate by approximately 12% and uses 11% less energy compared to the other methods. Increasing the cache hit rate can lead to a reduction in about 17% in response time for user requests. Our bi-level control approach offers a promising solution for leveraging the network's global visibility while balancing communication overhead (as energy consumption) against system performance. Additionally, the proposed method has the lowest cache eviction, around 19% lower than the lowest eviction of the other comparison methods. The eviction metric is a metric to evaluate the effectiveness of adaptive caching approach designed for transient data.

Current blockchain systems have high requirements on network connection and data transmission rate, for example, nodes have to receive the latest blocks in time to update the blockchain, nodes have to immediately broadcast the generated block to other nodes for consensus, which restricts the blockchain to run only on real-time connection networks, but the existence of delay tolerant networks poses a great challenge to the deployment of blockchain systems. To address this challenge, a novel blockchain transaction mechanism is proposed. First, the block structure is modified by adding a flag, and on this basis, the definition of the extrachain is proposed. Secondly, based on the blockchain transaction process, transaction verification and consensus algorithms on the extrachain are presented. Thirdly, both the extrachain selection algorithm and appending algorithm are proposed, so that the extrachain can be appended to the blockchain fairly and safely. Finally, an extrachain transmission scheme is presented to broadcast the blocks generated in the delayed network to the normal network. Theoretical analysis and simulation experiments further illustrate the efficiency of the proposed mechanism.