Modern Networks Research Articles

Learning Robust Shape Regularization for Generalizable Medical Image Segmentation.

Generalizable medical image segmentation enables models to generalize to unseen target domains under domain shift issues. Recent progress demonstrates that the shape of the segmentation objective, with its high consistency and robustness across domains, can serve as a reliable regularization to aid the model for better cross-domain performance, where existing methods typically seek a shared framework to render segmentation maps and shape prior concurrently. However, due to the inherent texture and style preference of modern deep neural networks, the edge or silhouette of the extracted shape will inevitably be undermined by those domain-specific texture and style interferences of medical images under domain shifts. To address this limitation, we devise a novel framework with a separation between the shape regularization and the segmentation map. Specifically, we first customize a novel whitening transform-based probabilistic shape regularization extractor namely WT-PSE to suppress undesirable domain-specific texture and style interferences, leading to more robust and high-quality shape representations. Second, we deliver a Wasserstein distance-guided knowledge distillation scheme to help the WT-PSE to achieve more flexible shape extraction during the inference phase. Finally, by incorporating domain knowledge of medical images, we propose a novel instance-domain whitening transform method to facilitate a more stable training process with improved performance. Experiments demonstrate the performance of our proposed method on both multi-domain and single-domain generalization.

Deep Learning-based Intelligent Fault Diagnosis for Power Distribution Networks

Power distribution networks with distributed generation (DG) face challenges in fault diagnosis due to the high uncertainty, randomness, and complexity introduced by DG integration. This study proposes a two-stage approach for fault location and identification in distribution networks with DG. First, an improved bald eagle search algorithm combined with the Dijkstra algorithm (D-IBES) is developed for fault location. Second, a fusion deep residual shrinkage network (FDRSN) is integrated with IBES and support vector machine (SVM) to form the FDRSN-IBS-SVM model for fault identification. Experimental results showed that the D-IBES algorithm achieved a CPU loss rate of 0.54% and an average time consumption of 1.70 seconds in complex scenarios, outperforming the original IBES algorithm. The FDRSN-IBS-SVM model attained high fault identification accuracy (99.05% and 98.54%) under different DG output power levels and maintained robustness (97.89% accuracy and 97.54% recall) under 5% Gaussian white noise. The proposed approach demonstrates superior performance compared to existing methods and provides a promising solution for intelligent fault diagnosis in modern distribution networks.

Architecture Research on Differentiated Lightning Protection of Distribution Network Considering Plateau Microtopography

Abstract Lightning is the main and severe threat to 10kV and 35kV medium-voltage distribution networks. It will cause equipment failures or damage, tripping and even power outages. With the construction of modern distribution networks, the requirements for power supply reliability are getting higher and higher. Therefore, lightning protection of distribution networks has received great attention. The key point is to find a cost-effective way to protect against lightning. This paper proposed an architecture for differentiated lightning protection of distribution networks considering plateau micro-topography. The main idea is to take multiple measures to obtain rich data, including a lightning positioning system, 3D laser scanning and digital elevation model. Differences in lightning parameters, line and tower parameters, topographic parameters are thought to bring different lightning tripping rates for each tower. Then tripping rate of each area element is obtained through meshing and discrete calculation. Total tripping rate is finally obtained by summation of all elements. The proposed architecture plans to provide a clue for differentiated lightning protection for the distribution network.

Empowering deep neural quantum states through efficient optimization

Computing the ground state of interacting quantum matter is a long-standing challenge, especially for complex two-dimensional systems. Recent developments have highlighted the potential of neural quantum states to solve the quantum many-body problem by encoding the many-body wavefunction into artificial neural networks. However, this method has faced the critical limitation that existing optimization algorithms are not suitable for training modern large-scale deep network architectures. Here, we introduce a minimum-step stochastic-reconfiguration optimization algorithm, which allows us to train deep neural quantum states with up to 106 parameters. We demonstrate our method for paradigmatic frustrated spin-1/2 models on square and triangular lattices, for which our trained deep networks approach machine precision and yield improved variational energies compared to existing results. Equipped with our optimization algorithm, we find numerical evidence for gapless quantum-spin-liquid phases in the considered models, an open question to date. We present a method that captures the emergent complexity in quantum many-body problems through the expressive power of large-scale artificial neural networks.

Machine Learning-based Enhanced Deep Packet Inspection for IP Packet Priority Classification with Differentiated Services Code Point for Advance Network Management

In modern networking, the efficient prioritization and classification of network traffic is paramount to ensure optimal network performance and optimization. This study presents an approach to enhance intelligent packet forwarding priority classification on Differentiated Services Code Point (DSCP), leveraging classifiers from machine learning algorithms for Deep Packet Inspection (DPI). The DSCP resides inside the Differentiated Services (DS) field of the Internet Protocol (IP) packet header in an OSI or TCP/IP model, which prioritizes different types of packets for forwarding to the router based on the attached payload. Similarly, DPI plays a crucial role in network management, enabling the identification of applications, services, and potential threats within the network traffic. In this study, various machine learning models, namely Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Decision Tree, Random Forest, Logistic Regression and ensemble models such as, XGBoost, AdaBoost were used to harness the capabilities of network packet classification based on DSCP. Detailed experimentation was conducted to evaluate their performance. The results show that AdaBoost demonstrated superior performance with an accuracy of around 89.91%, showcasing its ability to adapt the evolving network configurations and conditions while maintaining high classification accuracy on the IP packets. The random forest model also performed well, achieving an accuracy of 89.41%, making it a strong candidate for the DSCP classification in network transmission. This study has the potential to significantly improve how networks manage traffic, prioritize packets, and secure complex and dynamic network environments.

Simulation of Massive MIMO Antennas for Improved Link Throughput

Massive MIMO technology involves deploying multiple antennas on both ends of a communication link, which boosts spatial diversity and signal-to-noise ratio. By using multi-user MIMO transmission and techniques like spatial multiplexing and beamforming, it increases capacity, spectral efficiency, and data rates. Simulations conducted with tools like Octave software help engineers to optimize designs, predict system behavior, and assess the impact of antenna quantity on downlink throughput. This study addresses the increasing demand for high downlink data rates within modern communication networks. It explores the performance of Massive MIMO systems through simulations using Octave software, focusing on the quantity of antenna elements (M), and considering both spectral efficiency (SE) and data throughput. Results show higher M enhances both spectral efficiency and data rates. Efficiency dips at M=10, surges at M=40, and slowly, but steadily grows from M=50 to 100. Similar trends in data rates indicate M's impact on transmission performance. SE values range from 5.8442 bps/Hz (M = 10) to 24.1608 bps/Hz (M = 100), while data rates range from 116.8834 Mbps (M = 10) to 483.2158 Mbps (M = 100). These insights offer practical guidance for engineers and could facilitate the optimization of wireless communication setups and addressing the imperative for improved data rates. The study advances the understanding of Massive MIMO's potential, providing a roadmap for effectively leveraging its capabilities across diverse communication network scenarios.

COMPARATIVE ANALYSIS OF RANDOM FOREST AND ADABOOST LEARNING MODELS FOR THE CLASSIFICATION OF ATTACKS IN INTERNET OF THINGS

Attacks are actions that attempt to break one of the following properties of the computer system: confidentiality, integrity, and availability. The immense increment in the amount of internet applications and the appearance of modern networks has created the need for improved security mechanisms. Internet of Things (IoT) is a system that uses the Internet to facilitate communication between sensors and devices. Several approaches have been used to build attacks detection system in the past. This study built two ensemble models for the classification of attacks using Random Forest and Adaboost algorithms respectively. Feature importance was used for selecting promising attributes from the IoT intrusion dataset. Thereafter, the results of the classification models were evaluated and compared. The models were evaluated based on when feature selection technique was applied and without respectively. For Random Forest-based classification model with feature selection, 99.0% ,0.95,0.88,0.82, were obtained for accuracy, recall, f1-score, and precision respectively while without feature selection 69.0%,0.86,0.76,0.64 were obtained respectively. For Adaboost-based classification model with feature selection 99.0%.0.69,0.61,0.66 were obtained for accuracy, recall, f1-score and precision respectively. Without feature selection the Adaboost model recorded 58.0%,0.58,0.48,0.50 respectively. The results showed that both models achieved high rates with feature selection technique used, with Random Forest performing slightly better, both learning models showed promised performances in classifying attacks in IoT environments. This study concluded that the use of the chosen feature selection method helped improve the performances of the two ensembles in the classification of attacks in the IoT dataset.

Patterns of Stakeholder Relationships in the Implementation of the Tomira Partnership Program in Kulon Progo

TOMIRA stands for People's Owned Stores, which emerged as a reaction to the widespread growth of modern networked or franchised stores in the community, especially in Kulon Progo Regency. The development of modern network stores or franchises is considered a threat to the long-term existence of traditional markets, which in this case include cooperatives and micro, small and medium enterprises (MSMEs). TOMIRA is a partnership program between cooperatives, MSMEs, and large networked modern shop companies supported by the Kulon Progo Regency government in an effort to provide solutions for the sustainability of cooperatives and MSMEs in society. However, the TOMIRA partnership program still has various deficiencies, especially in terms of the roles and relationship patterns of the stakeholders involved. Therefore, the purpose of this study is to examine the pattern of relationships that occur between stakeholders involved in the implementation of the TOMIRA partnership program so that this program is expected to have a high success rate and be beneficial to various parties in the future.

Machine Learning-based fixed access Network Interface Congestion Prediction

Network interface congestion refers to a situation in which network interfaces or communication channels within a network experience high levels of traffic, leading to performance degradation or potential network failures. In modern networking, this issue remains relevant and can manifest in various contexts due to a surge in data traffic, Internet of Things (IoT) devices can generate substantial data, increased use of VPNs for remote work and when multiple users or applications concurrently access and transfer data in and out of the cloud. Efforts to forecast and address network interface bottlenecks are crucial to sustaining the performance and consistency of modern networks. In the PTCL fixed access network infrastructure, network blocking is a serious problem that necessities to be addressed. The main objective of this study is to overcome network congestion problems that occur in PTCL's fixed access network through the use of machine learning techniques. Despite having real-time monitoring tools such as SolarWinds, there remains a dearth of software capable of predicting imminent bottlenecks in the network interface. The primary aim is to build an analytical model which can predict potential blockages within 30 days and facilitate proactive management strategies for better administration. Top of Form The study classifies the most effective machine learning technique for this purpose and evaluates its performance against current methods. To achieve the goal, a diagnostic dataset was created, leveraging appropriate configurations, software tools, and hardware resources. Data was sourced from SolarWinds, consisting of information from over 23,000 interfaces across 11,000 access network nodes. This data was analyzed in PTCL's Data Warehouse, where a historical aggregated dataset was constructed. Exploratory data analysis was performed using Python libraries, including Matplotlib, NumPy, Pandas, and Seaborn. This investigation sets the foundation for proactive network management and enhanced user experiences within PTCL's fixed access network, addressing the critical essential for advanced predictive tools to combat congestion and enhance network competence.

Sentinel Shield: Leveraging ConvLSTM and Elephant Herd Optimization for Advanced Network Intrusion Detection

Given the escalating intricacy of network environments and the rising level of sophistication in cyber threats, there is an urgent requirement for resilient and effective network intrusion detection systems (NIDS). This document presents an innovative NIDS approach that utilizes Convolutional Long Short-Term Memory (ConvLSTM) networks and Elephant Herd Optimization (EHO) to achieve precise and timely intrusion detection. Our proposed model combines the strengths of ConvLSTM, which can effectively capture spatiotemporal dependencies in network traffic data, and EHO, which allow the model to focus on relevant information while filtering out noise. To achieve this, we first preprocess network traffic data into sequential form and use ConvLSTM layers to learn both spatial and temporal features. Subsequently, we introduce Elephant Herd Optimization that dynamically assigns different weights to different parts of the input data, emphasizing the regions most likely to contain malicious activity. To evaluate the effectiveness of our approach, we conducted extensive experiments on publicly available network intrusion CICIDS2017 Dataset. The experimental results demonstrate the efficacy of the proposed approach (Accuracy = 99.98%), underscoring its potential to revolutionize modern network intrusion detection and proactively safeguard digital assets.

Asynchronous Record Alignment of Network Flows for Incident Detection and Reconstruction

In today's interconnected digital landscape, the distribution of cyber threats presents a significant challenge to cyber security. Moreover, as of 2016, the amount of data in the world exceeds one zettabyte. Because of this, evidence-based network flow analytics is a critical component of modern network management and security. Problems such as anomalies in the network flow, cyber security incidents, alert generation, data pre-processing, network monitoring, network flow complexity, and data flow patterns become difficult to detect in massive network data flows. These specific problems can be addressed using Packet capture (PCAP). PCAP analysis is a standard network forensics process and investigation for assessing network behaviour and identifying anomalies. This work presents a method for analysing network flows for probable alignment of asynchronously recorded communications in heterogeneous networks. Using a proposed method for alignment, we can identify the relevant recordings aligned over two data streams for faster and more conclusive incident analysis. We use synthetic network incident scenarios for research experiments, detailing the generation of cyber event data and impact on cloud network traffic, followed by in-depth PCAP analysis. The automated cyber-attacks are simulated within a network infrastructure generating network flows in a PCAP format. Simulated cyber-attacks range from standard port scans, service scans, and specific scenarios like SQL injection, phishing, DoS or DDoS. We define analysis objectives and criteria for the in-depth PCAP analysis and alignment. The evidence gathered showcases valuable information about network data flow and its behaviour.

Forecasting VIX using Bayesian deep learning

AbstractRecently, deep learning techniques are gradually replacing traditional statistical and machine learning models as the first choice for price forecasting tasks. In this paper, we leverage probabilistic deep learning for inferring the volatility index VIX. We employ the probabilistic counterpart of WaveNet, Temporal Convolutional Network (TCN), and Transformers. We show that TCN outperforms all models with an RMSE around 0.189. In addition, it has been well known that modern neural networks provide inaccurate uncertainty estimates. For solving this problem, we use the standard deviation scaling to calibrate the networks. Furthermore, we found out that MNF with Gaussian prior outperforms Reparameterization Trick and Flipout models in terms of precision and uncertainty predictions. Finally, we claim that MNF with Cauchy and LogUniform prior distributions yield well-calibrated TCN, and Transformer and WaveNet networks being the former that best infer the VIX values for one and five-step-ahead forecasting, and the probabilistic Transformer model yields an adequate forecasting for the COVID-19 pandemic period.

ENHANCING NETWORK INTRUSION DETECTION WITH MOUNTAIN GAZELLE OPTIMIZER-BASED FEATURE SELECTION

Network intrusion detection systems (NIDS) play a critical role in safeguarding modern networks. Feature selection techniques are essential for optimizing NIDS performance by reducing dimensionality and enhancing classifier efficiency. This study proposes the Mountain Gazelle Optimizer (MGO), a novel nature-inspired metaheuristic, for effective feature selection within the context of network intrusion detection. MGO's performance is evaluated on the benchmark UNSW-NB15 dataset in conjunction with a diverse suite of classifiers: Support Vector Machines (SVM), K-Nearest Neighbors (KNN), Random Forest Classifiers (RFC), Decision Tree Classifiers (DTC), and Gaussian Naive Bayes (GNB). Our analysis investigates the impact of MGO-based feature selection on key NIDS performance metrics including accuracy, specificity, sensitivity, and computational runtime. Results demonstrate the potential of MGO to identify optimal feature subsets, leading to significant improvements in network intrusion detection performance across these varied algorithmic approaches. Additionally, we explore the comparative advantages of MGO-based feature selection for specific types of network attacks represented within the UNSW-NB15 dataset. Keywords: Intrusion Detection System (IDS), Mountain Gazelle Optimizer, Feature Selection, SVM, KNN, Naive Bayes, Classification, Optimization, Machine Learning.

Distributed SDN Security Frameworks: A Survey

Software-Defined Networks (SDN) have recently received much attention and deployment as a new technology that offers more flexibility and efficiency than traditional networks. SDN is a paradigm shift that revolutionizes traditional network design and makes future networks programmable, manageable, and affordable. The use of the SDN in modern networks provides much needed flexibility and transparency to organize and deploy network solutions. It is a new model that separates forwarding & controlling planes, and centralized architecture designed to increase network speed and programming capability. However, from the current security point of view, the SDN still has some problems, especially for the advanced persistent attacks such as the DDOS, the side channel attacks in Clouds, the SDN stack control plane saturation attacks, and the switch flow table exhaustion attacks. Also, the existing SDN-based security systems are constrained by a central framework that provides significant overheads for the control plane, leading to the breakdown of vital control links. In this paper, we will present the vulnerabilities and security threats in the SDN network, define the various approaches to solve these problems, and deploy the SDN securely in the production environments. We will survey existing research on distributed SDN security frameworks, there is a number of security frameworks and applications that have been proposed previously, and each one of them builds on a selection of the SDN characteristics.

Optimized Cross-Path Attacks via Adversarial Reconnaissance

While softwarization and virtualization technologies make modern communication networks appear easier to manage, they also introduce highly complex interactions within the networks that can cause unexpected security threats. In this work, we study a particular security threat due to the sharing of links between high-security paths and low-security paths, which enables a new type of DoS attacks, called cross-path attacks, that indirectly attack a set of targeted high-security paths (target paths) by congesting the shared links through a set of attacker-controlled low-security paths (attack paths). While the feasibility of such attacks has been recently demonstrated in the context of SDN, their potential performance impact has not been characterized. To this end, we develop an approach for designing an optimized cross-path attack under a constrained total attack rate, consisting of (i) novel reconnaissance algorithms that can provide consistent estimates of the locations and parameters of the shared links via network tomography, and (ii) efficient optimization methods to design the optimal allocation of attack rate over the attack paths to maximally degrade the performance of the target paths. The proposed attack has achieved a significantly larger performance impact than its non-optimized counterparts in extensive evaluations based on multiple network settings, signaling the importance of addressing such intelligent attacks in network design. For more detail, see the full paper [4].

Energy Efficiency Enhancement of Next Generation Mobile Systems- An Hybrid Approach

The surge in mobile users and online applications, particularly video streaming, is indeed fueling a significant increase in global data traffic. To accommodate this growth while ensuring fast data rates, it's crucial to optimize data transmission with limited resources. Balancing high data throughput with lower power consumption is essential for network efficiency without compromising Quality of Service (QoS). In our research, we focus on enhancing data transmission efficiency between base stations and multiple users. Our approach revolves around transmitting data only when there's a demand or explicit user request, a strategy aimed at conserving power. By doing so, we achieve substantial energy savings, leading to decreased energy consumption and improved energy efficiency. Central to our methodology is an optimized power control transmission mechanism. This ensures that data rates and QoS requirements are met while maximizing energy efficiency (EE). By dynamically adjusting power levels based on transmission needs, we strike an optimal balance between throughput and power consumption. Our findings underscore the significance of adaptive power control strategies in modern wireless networks. By intelligently managing resources and minimizing unnecessary transmissions, we pave the way for sustainable network growth and enhanced user experiences. DOI: https://doi.org/10.52783/jes.4768

Hawthorn with "homology of medicine and food": a review of anticancer effects and mechanisms.

Over the past few years, there has been a gradual increase in the incidence of cancer, affecting individuals at younger ages. With its refractory nature and substantial fatality rate, cancer presents a notable peril to human existence and wellbeing. Hawthorn, a medicinal food homology plant belonging to the Crataegus genus in the Rosaceae family, holds great value in various applications. Due to its long history of medicinal use, notable effects, and high safety profile, hawthorn has garnered considerable attention and plays a crucial role in cancer treatment. Through the integration of modern network pharmacology technology and traditional Chinese medicine (TCM), a range of anticancer active ingredients in hawthorn have been predicted, identified, and analyzed. Studies have shown that ingredients such as vitexin, isoorientin, ursolic acid, and maslinic acid, along with hawthorn extracts, can effectively modulate cancer-related signaling pathways and manifest anticancer properties via diverse mechanisms. This review employs network pharmacology to excavate the potential anticancer properties of hawthorn. By systematically integrating literature across databases such as PubMed and CNKI, the review explores the bioactive ingredients with anticancer effects, underlying mechanisms and pathways, the synergistic effects of drug combinations, advancements in novel drug delivery systems, and ongoing clinical trials concerning hawthorn's anticancer properties. Furthermore, the review highlights the preventive health benefits of hawthorn in cancer prevention, offering valuable insights for clinical cancer treatment and the development of TCM with anticancer properties that can be used for both medicinal and edible purposes.

Towards a quantum-safe 5G: Quantum Key Distribution in core networks

The Service-Based Architecture (SBA) of the fifth generation (5G) of cellular networks introduced great advancements in flexibility, efficiency, and performance of modern networks, thanks to the distribution and softwarization of core network entities called Network Functions (NFs). Due to the distributed nature of 5G systems, these NFs are usually located in different sites to provide better Quality of Service (QoS) or specific services, which poses a great challenge regarding security. However, even though the 3GPP standard identifies security requirements by leveraging Transport Layer Security (TLS) authentication and OAuth2.0 authorization, current experimental and commercial deployments often disregard these aspects for the sake of simplicity and performance. In fact, even though data and communication protection is crucial, the implementation of security protocols and mechanisms complicates the deployments and may reduce the efficiency of 5G systems. This paper presents a novel solution that implements TLS for authentication and encryption leveraging entanglement-based Quantum Key Distribution (QKD), ensuring that all the communications between NFs in the 5G core network are secured by QKD keys, and are therefore quantum-safe. Additionally, this paper presents a QKD key repository using the Network Repository Function (NRF), which stores and relays TLS session keys to authenticated consumer NFs to access producer NFs in a fast but quantum-safe process. The solution was validated in a real hardware environment, and obtained results demonstrate that the proposed solution enhances the security of the SBA communications in an efficient way, reducing an 85% the service request time compared to the traditional TLS-based procedure, as well as 29.96% fewer bytes transmitted throughout the process.

Product marketing mode of economic zone by genetic algorithm in economic chaos combination prediction

With the development of the social economy, people continue to innovate in marketing. The most important difference between product marketing and traditional marketing is that it transcends the boundaries of time and space. Due to the development of modern network technology, the product marketing model can operate efficiently and easily, increasing people’s purchasing power in time and space. This paper is a research on the economic zone product marketing model based on a genetic algorithm in predicting the economic chaos combination. It introduces in detail the research method of combining the prediction of economic chaotic combination with a standard genetic algorithm. It also specifically analyzes the impact of three marketing models: a complete online marketing model, an online and offline marketing model, and the introduction of a third-party implanted joint marketing model on the impact of product sales in the economic zone. The research results show that, compared with the first two marketing models, after the introduction of the third-party implanted joint marketing method, the sales of product A are better. Among them, the luxury goods of group c achieved the highest sales volume of 952 in November, while the daily necessities achieved the sales volume of 1607 in December.

Compact Model Training by Low-Rank Projection With Energy Transfer.

Low-rankness plays an important role in traditional machine learning but is not so popular in deep learning. Most previous low-rank network compression methods compress networks by approximating pretrained models and retraining. However, the optimal solution in the Euclidean space may be quite different from the one with low-rank constraint. A well-pretrained model is not a good initialization for the model with low-rank constraints. Thus, the performance of a low-rank compressed network degrades significantly. Compared with other network compression methods such as pruning, low-rank methods attract less attention in recent years. In this article, we devise a new training method, low-rank projection with energy transfer (LRPET), that trains low-rank compressed networks from scratch and achieves competitive performance. We propose to alternately perform stochastic gradient descent training and projection of each weight matrix onto the corresponding low-rank manifold. Compared to retraining on the compact model, this enables full utilization of model capacity since solution space is relaxed back to Euclidean space after projection. The matrix energy (the sum of squares of singular values) reduction caused by projection is compensated by energy transfer. We uniformly transfer the energy of the pruned singular values to the remaining ones. We theoretically show that energy transfer eases the trend of gradient vanishing caused by projection. In modern networks, a batch normalization (BN) layer can be merged into the previous convolution layer for inference, thereby influencing the optimal low-rank approximation (LRA) of the previous layer. We propose BN rectification to cut off its effect on the optimal LRA, which further improves the performance. Comprehensive experiments on CIFAR-10 and ImageNet have justified that our method is superior to other low-rank compression methods and also outperforms recent state-of-the-art pruning methods. For object detection and semantic segmentation, our method still achieves good compression results. In addition, we combine LRPET with quantization and hashing methods and achieve even better compression than the original single method. We further apply it in Transformer-based models to demonstrate its transferability. Our code is available at https://github.com/BZQLin/LRPET.