Computer Communications最新文献

Federated Learning (FL) has emerged as a powerful model for training collaborative machine learning (ML) models while maintaining the privacy of participants’ data. However, traditional FL methods can exhibit limitations such as increased communication overhead, vulnerability to poisoning attacks, and reliance on a central server, which can impede their practicality in certain IoT applications. In such cases, the necessity of a central server to oversee the learning process may be infeasible, particularly in situations with limited connectivity and energy management. To address these challenges, peer-to-peer FL (P2PFL) offers an alternative approach, providing greater adaptability by enabling participants to collaboratively train their own models alongside their peers. This paper introduces an original framework that combines P2PFL and Homomorphic Encryption (HE), enabling secure computations on encrypted data. Furthermore, a defense approach against poisoning attacks based on cosine similarity is introduced These techniques enable users to collectively learn while preserving data privacy and accounting for ideal energy optimization. The proposed approach has demonstrated superior performance metrics in terms of accuracy, F-scores, and loss when compared to other similar approaches. Furthermore, it offers robust privacy and security measures, leading to an enhanced security level, with improvements ranging from 5.5% to 14.6%. Moreover, we demonstrate that the proposed approach effectively reduces the communication overhead. The proposed approach resulted in impressive reductions in communication overhead ranging from 63.8% to 79.6%. The implementation of these security models was cumbersome, but we have made the code publicly available for your reference ¹.

Unmanned aerial vehicle (UAV)-enabled mobile edge computing (MEC) networks have recently been considered to be a support for ground MEC networks to enhance their computation capability. However, the line-of-sight (LOS) channels between the UAV and Internet of Things (IoT) devices can be interfered by various obstacles, such as trees and buildings, resulting in a considerable reduction in the capacity of MEC networks. To solve this issue, a system that combines multiple reconfigurable intelligence surfaces (RISs) with a UAV-enabled MEC network is proposed in this study. A UAV equipped with edge servers is treated as an aerial computing platform for IoT devices, and multi-RISs are utilized to simultaneously improve the communication quality between enhanced UAV and IoT devices. To maximize the sum computation bits of the system, a problem that jointly optimizes the time slot partition, local computation frequency, transmit power of the devices, UAV receive beamforming vectors, phase shifts of the RISs, and the trajectory of the UAV is formulated. The problem is a typical nonconvex optimization problem; therefore, we propose a recursive algorithm based on the successive convex approximation (SCA) and block coordinate descent (BCD) technology to find an approximate optimal solution. Simulation results demonstrate the effectiveness of the proposed algorithm compared with various benchmark schemes.

The exponential growth of Internet of Things (IoT) devices has triggered a substantial increase in cyber-attacks targeting these systems. Recent statistics show a surge of over 100 percent in such attacks, underscoring the urgent need for robust cybersecurity measures. When a cyber-attack breaches an IoT network, it initiates the dissemination of malware across the network. However, to counteract this threat, an immediate installation of a new patch becomes imperative. The time frame for developing and deploying the patch can vary significantly, contingent upon the specifics of the cyber-attack. This paper aims to address the challenge of pre-emptively mitigating cyber-attacks prior to the installation of a new patch. The main novelties of our work include: (1) A well-designed node-level model known as Susceptible, Infected High, Infected Low, Recover First, and Recover Complete $\left({\mathrm{SI}}_H{I}_L{R}_F{R}_C\right)$. It categorizes the infected node states into infected high and infected low, according to the categorization of infection states for IoT devices, to accelerate containment strategies for malware propagation and improve mitigation of cyber-attacks targeting IoT networks by incorporating immediate response within a restricted environment. (2) Development of an optimal immediate response strategy (IRS) by modeling and analyzing the associated optimal control problem. This model aims to enhance the containment of malware propagation across IoT networks by swiftly responding to cyber threats. Finally, several numerical analyses were performed to fully illustrate the main findings. In addition, a dataset has been constructed for experimental purposes to simulate real-world scenarios within IoT networks, particularly in smart home environments.

An extended connected dominating set (ECDS) in a wireless network with cooperative communication (CC) is a subset of nodes such that its induced subgraph is connected and each node outside the ECDS is covered by either one neighbor or several quasineighbors in the ECDS. Traditionality, the size of virtual backbone (VB) is the only factor considered in the problem of CDS construction. However, diameter is also an important factor to evaluate VB. In this paper we consider the problem of constructing quality ECDSs in unit disk graphs under CC with both of these two factors. We propose a two-phase centralized algorithm BD-ECDS to construct an ECDS for a given UDG with CC, which has a constant performance ratio (PR) and diameter. To obtain the PR of this two-phase centralized algorithm, we first give an upper bound of the EDS and use this upper bound to prove that the size of the ECDS under CC generated by the centralized algorithm is no greater than $120\left|ECD{S}_{opt}\right|-2$, where $ECD{S}_{opt}$ is the size of the minimum ECDS. Furthermore, our theoretical analysis and simulation results show that our algorithm BD-ECDS is superior to previous approaches.

With the widespread use of mobile communication and smart devices in the medical field, mobile healthcare has gained significant attention due to its ability to overcome geographical limitations and provide more efficient and high-quality medical services. In mobile healthcare, various instruments and wearable device data are collected, encrypted, and uploaded to the cloud, accessible to medical professionals, researchers, and insurance companies, among others. However, ensuring the security and privacy of healthcare data in the context of mobile networks has been a highly challenging issue. Certificateless signature schemes allow patients to conceal their respective privacy information for different sharing needs. Nevertheless, existing mobile healthcare data protection solutions suffer from costly certificate management and the inability to restrict signature verifiers. This paper proposes a certificateless designated verifier sanitizable signature for mobile healthcare scenarios, aiming to enhance the security and privacy of mobile healthcare data. This scheme enables the sanitization of sensitive data without the need for certificate management and allows for the specification of signature verifiers. This ensures the confidentiality of medical data, protects patient privacy, and prevents unauthorized access to healthcare data. Through security analysis and experimental comparisons, it is demonstrated that the proposed scheme is both efficient and effectively ensures data security and user privacy. Therefore, it is well-suited for privacy protection in mobile healthcare data.

The current trend in user services places an ever-growing demand for higher data rates, near-real-time latencies, and near-perfect quality of service. To meet such demands, fundamental changes were made to the traditional radio access network (RAN), introducing Open RAN (O-RAN). This new paradigm is based on a virtualized and intelligent RAN architecture. However, with the increased complexity of 5G applications, traditional application-specific placement techniques have reached a bottleneck. Our paper presents a Transfer Learning (TL) augmented Reinforcement Learning (RL) based networking slicing (NS) solution targeting more effective placement and improving downtime for prolonged slice deployments. To achieve this, we propose an approach based on creating a robust and dynamic repository of specialized RL agents and network slices geared towards popular user service types such as eMBB, URLLC, and mMTC. The proposed solution consists of a heuristic-controlled two-module-based ML Engine and a repository. The objective function is formulated to minimize the downtime incurred by the VNFs hosted on the commercial-off-the-shelf (COTS) servers. The performance of the proposed system is evaluated compared to traditional approaches using industry-standard 5G traffic datasets. The evaluation results show that the proposed solution consistently achieves lower downtime than the traditional algorithms.

With cyber threats evolving alongside technological progress, strengthening network resilience to combat security vulnerabilities is crucial. This research extends cyber-crime analysis with an innovative approach, utilizing data mining and machine learning to not only predict cyber incidents but also reinforce network robustness. We introduce a real-time data collection framework to provide up-to-date cyberattack data, addressing current research limitations. By analyzing collected attack data, we identified temporal correlations in attack volumes across consecutive time frames. Our predictive model, developed using advanced machine learning and deep learning techniques, forecasts the frequency of cyber-attacks within specific time windows, demonstrating over a 15% improvement in accuracy compared to conventional baseline models. The methodologies employed include the use of Recurrent Neural Networks (RNN) and Convolutional Neural Networks (CNN) for capturing complex patterns in time series data, and the integration of a sliding window technique to transform raw data into a format suitable for supervised learning. Our experiments evaluated the performance of various models, including ARIMA, Random Forest, Support Vector Regression, and K-Nearest Neighbors Regression, across multiple scenarios. Furthermore, we developed a Power BI platform for visualizing global cyber-attack trends, providing valuable insights for enhancing cybersecurity defences. Our research demonstrates that cyber incidents are not entirely random, and advanced AI tools can significantly enhance cybersecurity defences by analyzing patterns and trends from previous instances. This comprehensive approach not only improves prediction accuracy but also offers a robust framework for reducing the risk and impact of future cyber-crimes through enhanced detection and prediction capabilities.

This paper investigates the directional transmission scheme for airborne networks (ANs) with orthogonal time frequency space (OTFS) modulation, to cope with the degradations of communication performance due to the aircraft’s uncertain and high-speed mobilities. Besides showing better communication performance in high mobility scenarios such as AN, OTFS has also been verified for realizing the integrated sensing and communication (ISAC) systems. In this case, this paper proposes a sensing-assisted beam prediction method by exploiting echoes to predict the next locations of moving aircraft, solving the beam rendezvous problem at the transmitter. Besides, for the data detection problem at the receiver, this paper proposes a novel pilot placement scheme relying on the predicted delays and Dopplers, realizing accurate channel estimation with lower overheads. Simulation results show that the proposed OTFS-based directional transmission scheme can achieve reliable communication performance with a low bit error rate.

Providing cloud computing services leads to quick access of users to dynamic and distributed resources. Increasing demand has created challenges such as resource availability, privacy, and security to provide efficient services in cloud computing. Cloud environments contain various computing resources, and allocating a suitable node to process a request can improve the quality of service on a large scale. Load balancing is one of the strategies to improve service quality and resource utilization, which refers to the distribution of load among different nodes in a distributed system. The cloud application service broker is responsible for load balancing by choosing the appropriate geo-distributed datacenter to process the requests of each end user. Parameters such as transmission delay, network delay, processing time, number of servers, workload, and service cost can be considered to select a suitable datacenter in close proximity. To reduce the adverse effects of choosing a datacenter by a service broker, this paper presents Rank-based Load Balancing in Geo-Distributed datacenters (RLBGD) as an effective service broker strategy in cloud environments. RLBGD uses a weighted combination of several criteria such as processing time, number of servers, workload, processing speed, service cost, and response time for dynamic ranking and determining the appropriate datacenter. CloudAnalyst tool is used to simulate and analyze the performance of the proposed method. The results of experiments show the effectiveness of RLBGD in terms of metrics such as service cost and processing time in different scenarios.