Lossless fabrics are widely used in many production data centers, but they can give rise to issues such as head-of-line blocking and congestion spreading during network congestion, which can significantly degrade the performance of data center applications. Additionally, the latency of end-to-end solutions can lead to the buildup of switch queues. To address these challenges, this paper proposes a method called Priority Flow Control-Sensitive (PFC-S). PFC-S monitors buffer occupancy and traffic intensity, performs proactive rerouting, and prevents the impact of PFC congestion diffusion on victim flows. This approach helps maintain low buffer usage levels, thereby enabling control over tail latency. Initial evaluations demonstrate that PFC-S can reduce the average flow completion time and effectively prevent congestion spreading. Moreover, experimental results show that PFC-S provides better protection for victim flows compared to standard PFC, BFC, and HPCC methods.
The sixth generation (6G) of wireless communication envisions a diverse range of use cases, including high-mobility vehicular networks, non-terrestrial satellite links, and ultra-reliable low-latency communication scenarios. Conventional multicarrier waveforms, such as orthogonal frequency division multiplexing (OFDM), demonstrate constraints in highly dynamic and Doppler-rich settings, presenting difficulties in achieving 6G performance standards. In this context, orthogonal time frequency space (OTFS) modulation has emerged as a promising candidate due to its delay-Doppler domain processing and robustness to time-variant channel impairments. This paper provides a unified performance evaluation of OTFS for 6G, investigating its bit error rate (BER), energy efficiency (EE), and peak-to-average power ratio (PAPR) under realistic channel models. EE analysis reveals a strong dependency on circuit power consumption and hardware efficiency, where OTFS maintains superior EE across mobility regimes, particularly in vital vehicular-to-everything (V2X) links. Finally, PAPR evaluations highlight the trade-off between transmit efficiency and signal quality, with OTFS offering a manageable PAPR profile that can be optimized through pulse shaping and power control techniques. Evaluating BER, EE, and PAPR for OTFS under realistic 6G channel models provides a unified perspective on its practical deployment potential. These results underline OTFS as a promising modulation candidate for 6G systems, balancing reliability, energy sustainability, and hardware feasibility in high-speed vehicular networks.
Flying ad-hoc networks (FANETs) assist in high-risk tasks and monitor locations where human reachability is always at risk. FANETs have become one of the most appealing solutions for the Internet of Things in such scenarios. However, the default access mechanism in FANETs, binary exponential back-off, leads to increased collisions and energy utilization due to its lack of network adaptability. The proposed solution is a genetic fuzzy logic backoff (GFLB) algorithm, which determines the contention window size (CWS) based on the collision and success rate of the FANET. The genetic function optimizes the collision-success ratio by minimizing the fitness, while this optimized fitness value and retransmission attempts are then used to predict the CWS for subsequent transmission. A discrete chain Markov model with channel error is developed to evaluate the performance of the GFLB algorithm in terms of success, collisions, delay, and energy utilization under varying densities of unmanned aerial vehicles. The extensive simulation results demonstrate that the proposed GFLB algorithm significantly improves the efficiency and energy usage of end devices in FANETs compared to existing algorithms. The GFLB algorithm effectively reduces collision rates and energy consumption while enhancing the success rate and reducing delay in transmissions.
As efforts to reduce the environmental impact of energy production expand, renewable energy sources are becoming increasingly significant in the global energy balance. Over the anticipated 20-year life of a wind turbine, operation and maintenance (O&M) expenditures are predicted to account for 65%-90% of the overall investment cost, including inflation and crane charges. The higher estimate is based on 600-7500 kW machines in North America, while the lower estimate derives from the Danish fleet of 600 kW turbines. Reliability studies indicate that O&M costs contribute roughly 20%-25% of the levelized cost per kWh. These expenses strongly influence the profitability of wind farms and the competitiveness of wind turbines compared to other renewable energy options, highlighting significant potential for technological improvement. The profitability of a wind farm and the competitiveness of wind turbines compared to other green energy options are strongly influenced by O&M costs. Accurate wind speed and power generation prediction is therefore essential to improve efficiency and reduce investment costs. To address this, machine learning algorithms such as long-short term memory (LSTM), gated recurrent unit (GRU), artificial neural network (ANN), XGBoost, random forest (RF), and support vector machine (SVM) have been applied for forecasting wind speed and predicting power generation. Specifically, LSTM, GRU, and ANN are employed for wind speed forecasting, while XGBoost, RF, and SVM are used for electricity generation prediction. Results show that LSTM and GRU achieve lower root mean squared error than ANN in wind speed forecasting, while RF provides higher accuracy for power generation prediction compared to XGBoost and SVM. Overall, LSTM, GRU, and RF demonstrate strong performance in wind forecasting and power generation prediction.
Existing multi-path routing protocols can meet the service requirements of end-to-end delay and reliability between nodes in the Internet of Things, but they consume more energy. Constrained QoS routing is a new research method for routing protocols, which keeps up with the current trend. By considering the three QoS constraints of end-to-end delay, reliability, and energy consumption, we innovatively use the related techniques of mobile edge computing with machine learning automata to construct a sensor network as a multi-constrained optimal path model in this paper. At the same time, introducing the energy-aware node wake-up mechanism and the reward and punishment mechanism of learning automata, we propose an oriented mobile edge computing node energy-aware optimized QoS constrained routing algorithm. The algorithm can optimize the network energy consumption, extend the network life cycle, and accelerate the convergence by using learning automata. By experimental testing and comparison, the novel constrained Multi-QoS routing approach of energy-aware optimization based on learning automata for mobile edge computing (MQEN) proposed in this paper can reduce end-to-end delay, improve reliability, and decrease energy consumption.
With the widespread adoption of cloud computing, cloud storage technology has developed rapidly, but issues of data security and privacy protection have also become prominent. As a key means to ensure the security of cloud storage data, cloud auditing technology strengthens the data security defense for users by verifying data integrity and availability. In recent years, advancements in trusted execution environment (TEE) technology have brought higher security guarantees to cloud auditing. However, existing TEE-based multi-replica data auditing schemes, such as TEEMRDA (trusted execution environment-based multi-replica data audit), still have security vulnerabilities when facing specific attacks. This paper deeply analyzes the security risks of the TEEMRDA scheme, proposes an improved cloud data auditing scheme, comprehensively evaluating them from dimensions such as security and performance. The results show that the improved scheme performs excellently in resisting specific attacks and has significant advantages in performance. The scheme proposed by Hui Tian et al. (2025) exhibits potential security vulnerabilities under specific attack scenarios. The improved scheme can overcome the security problems and can be applied to unmanned aerial vehicles.
Massive multiple-input multiple-output (MIMO) systems are at the core of next-generation wireless communications because of their promise of high spectral efficiency, better signal quality, and power savings. However, the optimal beamforming design for such systems is an extremely compute-intensive problem, particularly when channel conditions vary, and the channel state information (CSI) is imperfect. This paper presents a reinforcement learning approach to beamforming that bypasses these limitations through an adaptive and data-driven policy learning paradigm for efficient transmission. The Q-learning-based reinforcement learning (RL) algorithm was run for 500 episodes and showed steady convergence, with the training loss decreasing from 0.87 to 0.04 and the validation loss from 0.90 to 0.05. Prominent performance parameters delivered by this approach constituted mean signal to interference plus noise ratio (SINR) is 22.9 dB, spectral efficiency 14.8 bits/s/Hz, and BER 3.1 x 10-4 at 15 dB signal to noise ratio (SNR). The SINR obtained could be improved by 50% and 22.5% with respect to maximum ratio transmission (MRT) and zero-forcing (ZF) techniques, respectively, while the power used was 9.6 W. Robustness testing with 5% CSI error and 100 Hz Doppler fading scenarios showed the very least deterioration in performance. This leads to the conclusion that RL in beamforming is a promising approach for real-time systems in view of future adaptive scalable MIMO deployments. This work can be extended in multi-user and multi-cell settings for more general applicability.
Swarm robotics aims to achieve robust collective behaviors through large numbers of relatively simple robots, but modeling and interpreting these emergent dynamics from real experimental data remains challenging. This work proposes an interpretable machine learning framework for modeling and analyzing swarm robot behaviors using a public IEEE DataPort dataset of swarm robotics experiments (eight robots, 200 time steps, and 1,600 labeled samples). We construct a feature-based representation of local interaction metrics (alignment, cohesion, separation, velocity, and position) and train a Random Forest classifier to recognize four behavioral phases: exploration, aggregation, formation, and foraging. The proposed classifier attains 98.12% overall accuracy and high per-class precision and recall, while feature importance and Shapley additive explanation analyses highlight alignment (31.44%) and cohesion (21.62%) as dominant behavioral drivers. Unsupervised clustering with KMeans and DBSCAN, supported by a Silhouette score of 0.2541 and an adjusted Rand index up to 0.69, reveals moderately separable latent structure consistent with the labeled phases. A Random Forest regressor further links local interaction features to global performance indicators, achieving high results on task-level outcomes. Our framework provides a unified, reproducible, and interpretable pipeline for real multi-robot data that combines classification, clustering, and regression. The results demonstrate that biologically inspired features can support accurate, explainable phase recognition and performance prediction, enabling data-driven design of swarm controllers for applications such as precision agriculture, search and rescue, and environmental monitoring.
Wireless sensor networks (WSNs) are widely applied in industrial scenarios for monitoring energy consumption, and their security should be guaranteed at all times. Unfortunately, they are not sufficiently secure because they have limited energy; therefore, they cannot afford to implement large-scale intrusion detection systems (IDSs). In addition, there is a lack of proper datasets related to industrial WSNs (IWSNs) to evaluate IDSs; thus, this study proposes an IWSN dataset based on OMNeT++. The proposed dataset is implemented using the low-energy adaptive clustering hierarchy (LEACH) protocol, and it validates the sinkhole and blackhole attack. We design a lightweight IDS named IWSN-SLEACH-IDS, which employs classification and regression trees (CART) to detect the sinkhole and blackhole attack, because the sinkhole and blackhole attack damage the factory, which relies on high-quality data. The defective product will cause millions of dollars of loss; therefore, CART achieves nearly 100% detection rate for the attack, and it consumes little additional energy, thus making it a suitable solution.
Trust evaluation for fifth generation (5G) and beyond 5G (B5G) network slicing has become increasingly critical due to the complexity of multi-tenant orchestration, dynamic resource allocation, and strict quality of service (QoS) requirements. However, conventional trust evaluation methods cannot capture the uncertainty coming from measurement noise, linguistic ambiguity, and dynamic network conditions. In this paper, we propose a fuzzy-based system for slice trust evaluation (FSSTE) designed for 5G/B5G network slicing environments. We implement two models: FSSTEM1 and FSSTEM2. FSSTEM1 considers three input parameters: QoS, security posture (SP), and isolation integrity (II), while FSSTEM2 extends this framework by incorporating resource reliability (RR) as a fourth parameter. Both models use interval type-2 fuzzy logic system to decide the slice trust (ST) output value. We evaluated the implemented models by simulations and found that when QoS, SP, II, and RR values increase, the ST value increases consistently. For FSSTEM1, when all parameters reach 0.9, the ST value is 0.868, which is suitable for mission-critical applications. For FSSTEM2, the RR provides 21% trust improvement under poor QoS conditions, which is suitable for safety-critical deployments. Parameter sensitivity analysis for FSSTEM1 shows that QoS and II have the same impact on ST (0.251). While SP has a stronger impact (0.257) compared with QoS and II. For FSSTEM2, the higher maximum footprint uncertainty of FSSTEM2 (0.249 vs. 0.190) suggests that four-parameters assessment involves greater uncertainty. FSSTEM2 is more complex than FSSTEM1 but provides better trust evaluation, especially in dynamic multi-tenant environments where RR changes.
Wireless mesh networks (WMNs) provide reliable and scalable wireless connectivity. They are also capable of dynamic data routing. However, mesh router placement in WMNs is a complex process and is classified as NP-hard problem. In this study, we present a WMN-PSOHCDGA hybrid system, which integrates particle swarm optimization (PSO), hill climbing (HC) and distributed genetic algorithm (DGA) to optimize mesh router placement. We compare two crossover methods: unimodal normal distribution crossover (UNDX) and simplex crossover (SPX) combined with six router replacement methods: constriction method (CM), random inertia weight method (RIWM), linearly decreasing inertia weight method (LDIWM), linearly decreasing V max method (LDVM), rational decreasing V max method (RDVM) and fast convergence rational decreasing V max method (FC-RDVM) for two island and stadium mesh client distributions. Simulation results for small-scale two-island distribution show that in all scenarios all mesh routers were connected and all clients were covered. While UNDX-LDIWM demonstrated better load balancing. For middle-scale two-island distribution, UNDX-LDIWM also achieved good load balancing although not all clients were covered, while both UNDX and SPX maintained full router connectivity. For stadium distribution, all mesh routers remained connected and CM, LDVM, and FC-RDVM achieved full client coverage, while CM and FC-RDVM demonstrated better load balancing.
Customers can access applications and storage through the cloud as a service. In the cloud, load balancing (LB) is crucial for distributing tasks between virtual machines (VMs). Utilizing a variety of LB techniques, users submit tasks to the cloud, which are distributed between VMs to speed up operations. LB speeds up VM operations. In the proposed work, two meta-heuristic algorithms, particle swarm optimization (PSO) and genetic algorithm (GA), are hybridized (hybridized GA and PSO (HGAPSO)) to achieve optimal LB. PSO is utilized to choose the global best VM, while the GA aids in selecting the best population from the initial population. The proposed work analyzed the different parameters, such as average processing time (APT), finishing time, reaction time, make-span, throughput, and cost. CloudSim tool with various levels of tasks is used for performance evaluation of different algorithms such as ant colony optimization (ACO), Harris Hawks optimization (HHO), and hybridization of HHO, GA, and PSO. The proposed HGAPSO algorithm demonstrates significant performance improvements over existing algorithms such as GA, PSO, HHO, ACO, and HHO–ACO. Experimental results show that HGAPSO achieves an 88.11% improvement in APT and a 70% reduction in average response time. Moreover, makespan is reduced by 62%, throughput is improved by 4.95%, and computational cost is lowered by 1.58%. These enhancements collectively confirm the efficiency of the proposed approach in achieving balanced load distribution and improved cloud performance.
In a distributed system framework, spatial crowdsourcing (SC) is a highly important area of research where task allocation to task executors (TEs) is an important step. Tasks are requested by a task provider and are allocated by an SC platform to TEs. However, TEs may submit the allocated task as late as possible, known as procrastination. Plenty of research works are available on task allocation in SC, whereas few research works are found that address procrastination. In a bipartite graph setting, a procrastination-aware scheduling is proposed. A recent work uses ChatGPT for procrastinating agents. Balanced distribution of tasks has not been addressed there. Recently, an algorithm was proposed that distributes tasks in a balanced manner in different slots to mitigate procrastination in SC. Here, we propose a quality-aware task allocation mechanism in an SC environment that combines a data science approach with a reinforcement learning-based approach. Once TEs are allocated tasks, we have proposed an AI-enabled (learning-the-variance) algorithm to distribute the tasks into slots with a more balanced distribution than any of the existing algorithms to mitigate procrastination. Our procrastination prevention mechanism outperforms existing methods, which is shown by extensive simulations. Analytically, it is shown that the proposed mechanism maintains a balanced distribution.
The GDPR impacts the design of information systems which process personal data, because it makes mandatory the adoption of the privacy-by-design and privacy-by-default principles. This compliance must be verified throughout the design cycle, so that it must be considered as early as possible in the cycle, when alternatives are not yet detailed in the overall design and just general directions of the projects may be available. A comparison between alternatives should be performed, which can only have a qualitative nature, but which involves numerous factors, so a panel of experts is needed to obtain a reliable result. In this paper, we propose a analytic hierarchy process-based evaluation approach to examine privacy-related features of alternative information system architectures in the early phases of the design cycle.
Background In recent years, cloud computing and cloud storage technologies have developed rapidly, and effectively verifying the integrity of cloud storage data has become the focus of researchers' attention.Objective To propose an improved scheme for overcoming the security problems in the scheme proposed by Imad El Ghoubach et al. in 2021. To discuss the application of the improved scheme in unmanned aerial vehicles (UAVs).Methods Through the cryptographic analysis of the original scheme and the improved scheme, it is pointed out that the original scheme has security problems, and the improved scheme is safe and correct. Through experimental simulation, the calculation cost of the original scheme and the improved scheme is compared.Results We point out that the original scheme cannot resist the collusion attack between the third-party auditor (TPA) and the cloud server provider (CSP), and TPA can forge the data block tags through the information provided by CSP. At the same time, the TPA can recover the user's data blocks using the evidence provided by the CSP.Conclusions The scheme proposed by Imad El Ghoubach et al. in 2021 is not safe. The improved scheme can overcome the security problems and can be applied to UAVs.
Network slicing (NS) is a technique that enables network operators to create multiple virtual networks, each customized for specific clients, services, or applications, while still utilizing a shared physical network infrastructure. Although this approach provides benefits in terms of resource usage and flexibility, it also introduces new security risks, particularly in the form of DDoS attacks. These attacks can be targeted at specific slices, causing disruptions to the services provided by those slices, which may impact multiple clients or applications that rely on those services. To mitigate the security risks posed by NS, the paper proposes an intrusion detection system that is designed to safeguard network slices from DDoS attacks. The proposed system relies on statistical methods that use joint entropy and dynamic thresholds to analyze network traffic in real time. Based on the findings of the testbed conducted for network slices, the proposed system exhibited a remarkable level of effectiveness in identifying DDoS attacks directed targeting a specific slice. The detection rate was recorded at 99%, and the delay rate was extremely low at 0.32 s. These results imply that the system can recognize and respond to attacks swiftly, which can aid in swiftly mitigating potential threats.
In recent years, battery electric vehicles (BEVs), which do not emit any CO2, have attracted considerable attention globally as a countermeasure against global warming. The plug-in BEV, a common BEV variant, is limited by the long charging time. The battery swapping electric vehicle (EV), however, allows for a relatively short fulfilling time in terms of electricity as its battery can be swapped for a pre-charged spare battery at a battery swapping station (BSS) in a few minutes. This paper presents an examination of the use of stored electricity as a power supply in emergencies. We propose both electricity migration for maintaining a sufficient amount of electricity among BSSs and methods that enable the reduction of the overheads incurred by EV users. The performance evaluations that validate the efficacy of the proposed methods are detailed, and the future research scope is indicated.
This work explores the use of generative adversarial networks (GANs) to tackle cyber-security challenges, including threat identification, anomaly detection, and mitigation strategies, particularly in complex systems and critical infrastructures like industrial control systems for energy production and distribution. GANs address a key obstacle in machine learning (ML)-based systems: the scarcity of quality data for training models capable of fully leveraging ML and deep learning in cyber-security applications, such as intrusion detection systems and malicious behaviour detectors. The study highlights GANs’ potential to enhance data augmentation by generating realistic synthetic network traffic flows. These flows simulate common cyber-attacks targeting operational technologies (OTs) and information technologies (ITs). A primary contribution of this research is the creation of a large, high-quality dataset of OT and IT network traffic, designed to improve the robustness of ML models used in cyber-defense systems. Additionally, the work includes statistical analyses to evaluate the reliability of GAN-based data augmentation, laying the foundation for further research. This approach promises significant advancements in developing resilient ML models capable of addressing evolving cyber-security threats.
Cybersecurity has increased on software-defined networks (SDNs) since its inception in the early 2010s. That is due to the SDN's centralized nature in controlling the network, human-centered generative methods, and transferring data in the internet of behavior. The widespread usage of Internet of Things (IoT) networks, integrating SDNs with IoT, the intelligent industrial revolution, and the inclusion of the human-centric industrial revolution in the infrastructure of major industrial fields all were reasons to focus on creating effective attack detection systems intrusion detection system (IDS) for these industrial IoT (IIoT) networks. In this paper, we present an IDS we call SPAARC_DDSDN_IIoT. It is a four-layer industrial IoT architecture weaponized by security devices at each layer and uses the SDN's centralized approach. First is an application layer-based prediction approach called SPAARC. SPAARC is a split-point algorithm combined with an attribute-reduced classifier (SPAARC). It is applied to the proposed IDS-based SDN approach in the IIoT. SPAARC is a decision tree algorithm, and intrusion detection will be performed on the leaves of the resulting tree. Two datasets were used for the experiment: the DDoS_SDN and the XIIoT_ID. The firefly algorithm, as a swarm intelligence based on the behavior of fireflies, was used for feature selection. SPAARC achieved a notable accuracy of 99.9962% and 99.991%, surpassing all the other machine learning algorithms tested. SPAARC also achieved a mean absolute error of 0 and 0.008, a root mean squared error of 0.0062 and 0.0016, and a perfect score on both datasets for the remaining metrics.
In a two-dimensional photonic crystal T-shaped waveguide with a passive switching function, the routing characteristics depending on the input light amplitude and the wavelength are numerically demonstrated by using the frequency-dependent finite-difference time domain method. For typical input electric field amplitudes ( E 0 = 0.1, 0.5, 1.0, and 1.5 [V/m]), the distribution ratio to the outputs is evaluated from the electric field profile. Additionally, the cross-correlation coefficient and normalized power spectra are considered. For the add/drop circuit in wavelength division multiplexing (WDM) optical network, it has been determined that the add/drop signal wavelength can be tuned by engineering the duplexer at the branching point with specific pillar radii.
