Network Flow Research Articles

Evaluating the Performance and Challenges of Machine Learning Models in Network Anomaly Detection

The application of machine learning algorithms for anomaly detection in network traffic data is examined in this study. Using a collection of network flow records that includes attributes such as IP addresses, ports, protocols, and timestamps, the study makes use of correlation heatmaps, box plots, and data visualization to identify trends in numerical characteristics. After preprocessing, which includes timestamp conversion to Unix format, three machine learning models Support Vector Machine (SVM), Gaussian Naive Bayes, and Random Forest are used for anomaly identification. The Random Forest Classifier outperforms SVM and Naive Bayes classifiers with better precision and recall for anomaly diagnosis, achieving an accuracy of 87%. Confusion matrices and classification reports are used to evaluate the models, and they show that the Random Forest Classifier performs better than the other models in identifying abnormalities in network traffic. These results provide significant value to the field of cybersecurity by highlighting the effectiveness of machine learning models specifically, the Random Forest Classifier in boosting anomaly detection capacities for network environment security.

Design of Radial Distribution System to Study Load-Flow and Short Circuit Analysis Using ETAP Software

This work's analysis provides information on fault analysis and load flow studies, which help to understand how radial distribution networks operate both in normal and emergency circumstances. The research explores optimization methods, describing capacitor placement, transformer size, and feeder reconfiguration in order to lower losses and increase system efficacy. Moreover, the work looks at reliability-focused maintenance techniques, stressing the need of predictive maintenance and condition monitoring in raising the dependability of radial distribution networks. Modern electrical power networks depend heavily on their radial distribution system, which effectively and dependable distributes electricity to customers. Many radial distribution system-related topics are covered in detail on this website, including analytical methods, optimization processes, and dependability enhancement strategies. Power flow, voltage control, and losses in radial networks are investigated in this paper under a range of operational situations. The last and crucial stage of a power system, the electrical distribution system makes sure that customers have a steady supply of electricity. Regretfully, because it is often neglected and exposed to environmental risks, our distribution network has a number of problems. Our distribution feeder is rigorously analyzed for load flow and short-circuits in order to overcome these challenges and find and fix problems. The main topics of this study are the load flow and short-circuit analysis of a radial distribution feeder, which is mostly constructed with (ETAP). The final and crucial part of a power system, the electrical distribution system ensures that consumers receive an uninterrupted and steady supply of electricity. Our distribution feeder undergoes a rigorous and cautious process to address these issues front on because of the uncontrollable nature and susceptibility of our network to environmental hazards. This paper delves deeply into the issues of short-circuiting and load flow in radial distribution feeders. Both symmetrical and asymmetrical faults are thoroughly investigated by mathematical computations and ETAP simulations; the results of the software simulations are contrasted with those of the mathematical analysis. These discoveries will definitely influence the development and management of our system and point us in the direction of workable and practical solutions.

Particulate and water-mobilizable phosphorus from a watershed with a plain river network contributes equal amounts of algal available phosphorus to its downstream lake

This research was designed to estimate the contributions of phosphorus (P) in different factions from an upstream plain river network to algal growth in a downstream shallow eutrophic lake, Taihu Lake, in China. During three flow regimes, the P fractions in multiple phases (particulate, colloidal and dissolved phases) and their algal availabilities were assessed via bioassays with Dolichospermum flos-aquae as the test organism. The P partitioning patterns among multiple phases were strongly affected by the concentration of total suspended solids (TSS) that changed with the river flow regime, with stronger disturbance of sediments at lower water levels (low flow) and weaker disturbance of sediments at higher water levels (high flow) in the plain river network. The median TSS concentration across the river network decreased from 157.4 mg/L during low flow to 31.8 mg/L during high flow, and the median particulate P concentration decreased from 0.132 mg/L to 0.093 mg/L. The particulate P contributed equally to the amount of algal available P (AAP) as did the water-mobilizable P (colloidal plus dissolved phase) in the rivers flowing into Taihu Lake. The annual average concentrations of particulate algal available P (P-AAP), colloidal algal available P (C-AAP) and dissolved algal available P (D-AAP) were estimated to be 0.032 mg/L, 0.012 mg/L and 0.019 mg/L, respectively, during 2012–2018, accounting for 50.8 %, 19.0 % and 30.2 %, respectively, of the total AAP. At the watershed scale, controlling P drainage from downstream urbanized areas should be emphasized. Additionally, controlling sediment resuspension or reducing the TSS concentration in the inflowing rivers is important for decreasing the particulate P flux to downstream lakes.

Кількісне оцінювання функціональної ефективності мережевих структур за аналізом чутливості оптимізаційної моделі до зовнішніх впливів

The real network organizational structure is constantly under the influence of external influences, which can cause unexpected consequences due to the cascading spread of relevant disturbances directed at critically important parameters. Parametric sensitivity analysis is the latest computational procedure, the results of which provide the researcher with an idea and quantitative assessment of the appropriate response of the structure in the form of an analytical platform as a composition and interaction of scenario analysis, cascade process development and optimization modeling technology. The main task is timely adaptation to expected changes in business analytics, oriented to optimal resources distribution. The generalized minimum cost problem is a typical example of widespread problems of optimal distribution and use of limited, in particular, energy resources, of any nature and purpose, using the scientific management and analysis of decisions, in its network model attention is paid to the potentials of nodes. The maximum flow problem is an example of the active and effective application of methods and models of flows optimization, where the processes and objects used a network organization, the energy systems and complexes investigated, where arc parameters are critical in its model. In the article, for these classic optimization problems, mathematical and spreadsheet models were built on concrete examples. Direct and dual mathematical programming problems were solved, the following stages were carried out - the organization of a computer experiment and the construction of an analytical platform for evaluating the behavior of the network structure under the influence of external influences and disturbances as their consequences. To take into account the specific conditions of real objects with a network structure at the stage of modification, the model supplemented with appropriate restrictions and correction of input data sets. The obtained results should be useful for planning and management personnel and top-management for discussion and decision-making regarding, for example, the optimal placement of energy equipment, the weighted distribution of energy flows in the "source-sink" system or the appointment of executors for work, automated project management using project networks, etc. Keywords: optimization modeling, maximum network flow, minimum cost flows, network organizational structures, decision making, spreadsheet modeling and analytics.

Transmural Flow Upregulates PD-L1 Expression in Microvascular Networks.

Endothelial programmed death-ligand 1 (PD-L1) expression is higher in tumors than in normal tissues. Also, tumoral vasculatures tend to be leakier than normal vessels leading to a higher trans-endothelial or transmural fluid flow. However, it is not clear whether such elevated transmural flow can control endothelial PD-L1 expression. Here, a new microfluidic device is developed to investigate the relationship between transmural flow and PD-L1 expression in microvascular networks (MVNs). After treating the MVNs with transmural flow for 24h, the expression of PD-L1 in endothelial cells is upregulated. Additionally, CD8 T cell activation by phytohemagglutinin (PHA) is suppressed when cultured in the MVNs pre-conditioned with transmural flow. Moreover, transmural flow is able to further increase PD-L1 expression in the vessels formed in the tumor microenvironment. Finally, by utilizing blocking antibodies and knock-out assays, it is found that transmural flow-driven PD-L1 upregulation is controlled by integrin αVβ3. Overall, this study provides a new biophysical explanation for high PD-L1 expression in tumoral vasculatures.

Quantum deep learning-based anomaly detection for enhanced network security

Identifying and mitigating aberrant activities within the network traffic is important to prevent adverse consequences caused by cyber security incidents, which have been increasing significantly in recent times. Existing research mainly focuses on classical machine learning and deep learning-based approaches for detecting such attacks. However, exploiting the power of quantum deep learning to process complex correlation of features for anomaly detection is not well explored. Hence, in this paper, we investigate quantum machine learning and quantum deep learning-based anomaly detection methodologies to accurately detect network attacks. In particular, we propose three novel quantum auto-encoder-based anomaly detection frameworks. Our primary aim is to create hybrid models that leverage the strengths of both quantum and deep learning methodologies for efficient anomaly recognition. The three frameworks are formed by integrating the quantum autoencoder with a quantum one-class support vector machine, a quantum random forest, and a quantum k-nearest neighbor approach. The anomaly detection capability of the frameworks is evaluated using benchmark datasets comprising computer and Internet of Things network flows. Our evaluation demonstrates that all three frameworks have a high potential to detect the network traffic anomalies accurately, while the framework that integrates the quantum autoencoder with the quantum k-nearest neighbor yields the highest accuracy. This demonstrates the promising potential for the development of quantum frameworks for anomaly detection, underscoring their relevance for future advancements in network security.

Applying self-supervised learning to network intrusion detection for network flows with graph neural network

Graph Neural Networks (GNNs) have garnered intensive attention for Network Intrusion Detection System (NIDS) due to their suitability for representing the network traffic flows. However, most present GNN-based methods for NIDS are supervised or semi-supervised. Network flows need to be manually annotated as supervisory labels, a process that is time-consuming or even impossible, making NIDS difficult to adapt to potentially complex attacks, especially in large-scale real-world scenarios. The existing GNN-based self-supervised methods focus on the binary classification of network flow as benign or not, and thus fail to reveal the types of attack in practice. This paper studies the application of GNNs to identify the specific types of network flows in an unsupervised manner. We first design an encoder to obtain graph embedding, that introduces the graph attention mechanism and considers the edge information as the only essential factor. Then, a self-supervised method based on graph contrastive learning is proposed. The method samples center nodes, and for each center node, generates subgraph by it and its direct neighbor nodes, and corresponding contrastive subgraph from the interpolated graph, and finally constructs positive and negative samples from subgraphs. Furthermore, a structured contrastive loss function based on edge features and graph local topology is introduced. To the best of our knowledge, it is the first GNN-based self-supervised method for the multiclass classification of network flows in NIDS. Detailed experiments conducted on four real-world datasets (NF-Bot-IoT, NF-Bot-IoT-v2, NF-CSE-CIC-IDS2018, and NF-CSE-CIC-IDS2018-v2) systematically compare our model with the state-of-the-art supervised and self-supervised models, illustrating the considerable potential of our method. Our code is accessible through https://github.com/renj-xu/NEGSC.

A framework for detecting zero-day exploits in network flows

Zero-day attack detection solutions aim to proactively identify unknown threats targeting valuable assets within a given system. While many Intrusion Detection System (IDS) solutions leverage learning techniques to build novel attack detection systems, they often focus on enhancing accuracy for specific attack types, overlooking the potential for multiple attack scenarios. Therefore, we introduce a novel framework for detecting zero-day attacks that evade current detection systems. Our framework enhances attack identification and qualification through a hybrid learning approach, where supervised learning ensures detection of known attacks and unsupervised learning. It encompasses intrusion detection phases from data collection to new attack class detection by identifying anomalies in real-time network flow data. Unsupervised learning, which involves grouping similar data points into clusters, establishes minimum distances within these clusters. This process triggers cluster division when certain thresholds are reached. Finally, an online supervised learning process validates our approach’s effectiveness in identifying anomalies associated with zero-day attack flows.This approach significantly reduces the False Detection Rate (FDR) without solely focusing on optimizing machine learning (ML) and deep learning (DL) algorithms hyper-parameters. We evaluated our framework on two datasets: one from a real industrial context at IBM and the NSL-KDD dataset. The results demonstrate our framework’s ability to detect anomalies in previously zero-day attack targets. On average, we identified 71 anomalous flows per target, achieving an overall average online learning accuracy of 98.4% for the IBM dataset and 96.6% for the NSL-KDD dataset, thereby validating the detection of these new attack scenarios.

Algorithmic approach to discrete fracture network flow modeling in consideration of realistic connections in large-scale fracture networks

Analyzing rock mass seepage using the discrete fracture network (DFN) flow model poses challenges when dealing with complex fracture networks. This paper presents a novel DFN flow model that incorporates the actual connections of large-scale fractures. Notably, this model efficiently manages over 20,000 fractures without necessitating adjustments to the DFN geometry. All geometric analyses, such as identifying connected fractures, dividing the two-dimensional domain into closed loops, triangulating arbitrary loops, and refining triangular elements, are fully automated. The analysis processes are comprehensively introduced, and core algorithms, along with their pseudo-codes, are outlined and explained to assist readers in their programming endeavors. The accuracy of geometric analyses is validated through topological graphs representing the connection relationships between fractures. In practical application, the proposed model is employed to assess the water-sealing effectiveness of an underground storage cavern project. The analysis results indicate that the existing design scheme can effectively prevent the stored oil from leaking in the presence of both dense and sparse fractures. Furthermore, following extensive modification and optimization, the scale and precision of model computation suggest that the proposed model and developed codes can meet the requirements of engineering applications.

A Finite State Machine Model for Business Flow in Active Distribution Networks Considering the Multi-flow Fusion Process

A Finite State Machine Model for Business Flow in Active Distribution Networks Considering the Multi-flow Fusion Process

Distributionally robust disaster relief planning under the Wasserstein set

We study a two-stage natural disaster management problem modeled as a stochastic program, where the first stage consists of a facility location problem, deciding where to open facilities and pre-allocate resources such as medical and food kits, and the second stage is a fixed-charge transportation problem, routing resources to affected areas after observing a disaster. Our model has binary variables present in both stages. Due to the lack of data, classical stochastic programming approaches may be ill-suited, and we propose a two-stage distributionally robust formulation with a Wasserstein ambiguity set, where we consider distributions consistent with historical data and a tunable parameter to control the level of risk aversion. We develop a tailored column-and-constraint generation (CCG) algorithm to solve an extensive reformulation, where scenarios are iteratively generated. We handle the presence of binary variables in the second stage by leveraging the structure of our support set and second-stage problem, and provide conditions under which the optimal value of the latter is concave with respect to the intensity of the disaster, leading to an efficient scenario generation procedure. We also show that our results extend to the case where the second stage is a fixed-charge network flow problem. We perform extensive computational experiments demonstrating the computational advantage of our method over classical CCG implementations on synthetic instances, and illustrate the benefits of our approach on a popular case study from the literature of hurricane threats on the Gulf of Mexico in the United States.

A solution for abstract decision problems based on maximum flow value

An abstract decision problem is an ordered pair where the first component is a nonempty and finite set of alternatives from which a society has to make a choice and the second component is an irreflexive relation on that set representing a dominance relation. A crucial problem is to find a reasonable solution that allows to select, for any given abstract decision problem, some of the alternatives. A variety of solutions have been proposed over the years. In this paper we propose a new solution, called maximum flow value set, that naturally stems from the work by Bubboloni and Gori (The flow network method, Social Choice and Welfare 51, pp.621-656, 2018) and that is based on the concept of maximum flow value in a digraph. We analyze its properties and its relation with other solutions such as the core, the admissible set, the uncovered set, the Copeland set and the generalized stable set. We also show that the maximum flow value set allows to define a new Condorcet social choice correspondence strictly related to the Copeland social choice correspondence and fulfilling lots of desirable properties.

An Engine-Level Safety Assessment Approach of Sustainable Aviation Fuel Based on a Multi-Fidelity Aerodynamic Model

Safety is essential for sustainable aviation fuels (SAFs). However, evaluating SAFs’ impacts on aero-engine safety is challenging because it involves multiple space scales and the strongly coupled relationships of aero-engine components. Aiming at addressing this problem, a model-based approach is proposed to establish the relationship between the fuel-level physical properties and engine-level safety parameters. Firstly, a unified modeling criterion is proposed to consider the interrelations of aero-engine components. Under this criterion, aero-engine secondary air system (SAS) components are included in SAF safety assessment, since they have non-neglectable influences on aero-engine safety. Secondly, this paper proposes a surrogate-based iteration strategy to embed the combustor’s high-dimensional computational fluid dynamics (CFD) model into the aero-engine flow network model. Then, the proposed model-based safety assessment approach is applied to a Fischer–Tropsch hydro-processed synthesized paraffinic kerosine (F-T SPK) safety assessment case. The effects of fuel flow and blending ratio are considered. The results indicate the necessity to evaluate SAFs’ safety at the aero-engine level and consider the influences of SAS components. The proposed model-based approach may provide a preliminary screening before SAFs’ certification tests. This convenience may be beneficial for reducing the cost and accelerating SAFs’ application.

Construction and Change Analysis of Water Ecosystem Service Flow Networks in the Xiangjiang River Basin (XRB)

Clearing and successfully characterizing ecosystem service flow paths has become a key bottleneck restricting in-depth research on the supply and demand relationships of ecosystem services. At present, although some explorations have been performed using water ecosystem services as a pioneer, the nature of its network and the fact that ecological base flow needs to be eliminated have been ignored. This study used InVEST and network models to consider ecological base flow, quantifying the supply, demand, and flow paths of freshwater ecosystem services in the Xiangjiang River Basin. The results showed that the overall distribution of the water supply in the Xiangjiang River Basin from 2000 to 2020 shows a pattern of higher supply in the south and lower supply in the north. The distribution of water demand shows higher levels in the north and lower levels in the south. The network density remains at its maximum level. The results of this study have provided a scientific basis for water resource management in river basins and improving ecological compensation mechanisms.

Study on a Strong Polymer Gel by the Addition of Micron Graphite Oxide Powder and Its Plugging of Fracture

It is difficult to plug the fracture water channeling of a fractured low-permeability reservoir during water flooding by using the conventional acrylamide polymer gel due to its weak mechanical properties. For this problem, micron graphite powder is added to enhance the comprehensive properties of the acrylamide polymer gel, which can improve the plugging effect of fracture water channeling. The chemical principle of this process is that the hydroxyl and carboxyl groups of the layered micron graphite powder can undergo physicochemical interactions with the amide groups of the polyacrylamide molecule chain. As a rigid structure, the graphite powder can support the flexible skeleton of the original polyacrylamide molecule chain. Through the synergy of the rigid and flexible structures, the viscoelasticity, thermal stability, tensile performance, and plugging ability of the new-type gel can be significantly enhanced. Compared with a single acrylamide gel, after adding 3000 mg/L of micrometer-sized graphite powder, the elastic modulus, the viscous modulus, the phase transition temperature, the breakthrough pressure gradient, the elongation at break, and the tensile stress of the acrylamide gel are all greatly improved. After adding the graphite powder to the polyacrylamide gel, the fracture water channeling can be effectively plugged. The characteristics of the networked water flow channel are obvious during the injected water break through the gel in the fracture. The breakthrough pressure of water flooding is high. The experimental results are an attempt to develop a new gel material for the water plugging of a fractured low-permeability reservoir.

Analysis of Intercity Transportation Network Efficiency Using Flow-Weighted Time Circuity: A Case Study of Seven Major City Clusters in China

Enhancing the efficiency of intercity transportation networks is crucial for sustainable regional transport development, significantly impacting travel behaviors and energy consumption. The transportation infrastructure within the city cluster is rapidly developing to accommodate the increasing traffic demand, necessitating substantial investments. It is imperative to investigate the effectiveness of intercity traffic within urban clusters, to evaluate the influence of transportation infrastructure enhancements on regional traffic efficiency. Circuity is a conventional metric used to assess the efficiency of transportation networks, primarily emphasizing distance, while overlooking factors such as travel time and traffic flow. In this study, the concept of circuity has been redefined in terms of travel time and has been referred to as the transportation network travel speed. Subsequently, the amalgamation of travel speed within the transportation network and traffic flow culminates in the proposition of Flow-Weighted Time Circuity (FWTC). Real-time intercity navigation data, offering accurate travel time estimations, are utilized to analyze the spatial distribution of intercity transport efficiency in the seven major city clusters of China, via both automobile and train modes of transportation. The results indicate that (1) as the travel distance extends, the speed of transportation within the network typically increases, albeit with increasing fluctuations, especially in the case of intercity train travel; (2) concerning the efficiency of intercity automobile travel, most city clusters demonstrate satisfactory performance, with the exception of the Guanzhong Plain. The Yangtze River Delta and Beijing–Tianjin–Heibei regions stand out for their superior performance. In terms of intercity train efficiency, the Yangtze River Delta, Beijing–Tianjin–Heibei, and Mid-Yangtze River regions exhibit higher levels of efficiency in intercity train transportation, while the Guanzhong Plain city cluster falls behind in this aspect. On the whole, the efficiency of intercity travel using automobiles surpasses that of train travel, indicating a pressing need for improvement in the latter.

Optimizing LPG production under quality specifications by preprocessing MILP with Pareto dominance and DEA

Liquified Petroleum Gas (LPG) is an oil refinery product that must adhere to quality specifications with respect to certain impurities. Refineries apply an LPG purification process that consists of a flow network with several process units (PUs). Current methods focus on optimising the performance of each PU separately; there exists no known approach for identifying the whole process optimum. In this paper, we present an approach for optimising the LPG purification process as a whole. We utilise operational scenarios to model the non-linear transformations of each PU. These scenarios enable us to devise a Mixed Integer Linear Program (MILP) that minimises energy consumption. The obtained solution is an approximation of the optimum but offers actionable support to refinery engineers. To enable the applicability of our approach at large scale, we propose two filters based on Pareto dominance and Data Envelopment Analysis (DEA) to identify the Pareto optimal and the Best Practice set of scenarios, respectively. By filtering out the rest, we reduce the solution space of the corresponding MILP. Further, we provide computational evidence that the application of these filters vastly improves solution time and enables applicability under real production conditions.

Analysis of spatial characteristics and influencing factors of the flow network of highly educated talents from national and local perspective

Based on dynamic monitoring data on China’s population, by using complex networks, spatial analysis and mathematical measurement, this study reveals the spatial characteristics and influencing factors of the network of flows of highly educated talents in the Yangtze River Delta region from the national and local perspectives. In the two perspectives, the network has strong isomorphism and certain differences. The in-flow of highly educated talents from cities with high administrative levels and more developed economies to Shanghai constitutes the core of the entire network. From a national perspective, highly educated talents tend to converge to the Yangtze River Delta region. From a local perspective, it was found that these talents cluster towards a limited number of cities in the region. From both perspectives, the flow network has developed into a “core-periphery” progressive hierarchical structure, with Shanghai becoming the sole core city. There is little difference in the influencing factors of talent mobility from both macro and meso perspectives. Highly educated talents would frequently flow between cities with strong economic development levels, and cities with high education level, scientific and technological level, complete infrastructure, and good aesthetics. However, geographical distance still plays a hindering role in the flow of highly educated talents, and factors such as cultural identity, institutional, and social modality differences among regions also have a certain effect on the flow of these talents.

Aggregation formulation for on‐site multidepot vehicle scheduling scenario

AbstractThe multidepot vehicle scheduling problem (MDVSP) is a fundamental public transport challenge. To address the large‐scale model and inherent solution symmetry associated with the traditional trip‐to‐trip connection‐based approach for MDVSP, a new trip‐to‐route (T2R) connection‐based approach is proposed. Considering real‐world problem characteristics with numerous trips sharing common origin–destination stations and travel times on one route, this approach aggregates same vehicle possible trip sequences into a T2R connection. Two time‐space network aggregation (TSNA) flow formulation versions, route pair‐based TSNA and station pair‐based TSNA, were constructed. Furthermore, TSNA equivalence under any given decomposition strategy, including first‐in‐first‐out, with the multicommodity network flow (MCNF) model was demonstrated. Given the favorable separable TSNA structure, an alternating direction method of multipliers (ADMM)‐based procedure is proposed to decompose the MDVSP into multiple subproblems that can be linearized and readily solved using commercial solvers. The quality of the solutions was assessed using lower bounds obtained from the Lagrangian relaxation problem. The effectiveness and superiority of the proposed MDVSP models and algorithms were subsequently confirmed using random data sets and real‐world instances.

Passivity-based control of fluid flow networks with capacitance

Fluid flow networks (FFNs) widely exist in industrial energy systems tightly related with the industrial production and residents living, such as district heating networks (DHN), mine-ventilation systems and gas transportation systems. The FFN operation is to drive the flowrates and pressures to their desired values, and the flowrate-pressure coordinated control of FFN is crucial in providing safe, stable and efficient operation. Although fluid capacitance has the same importance as fluid resistance and fluid inductance in giving dynamical characteristics of FFN, fluid capacitance is still not fully considered in the flowrate and pressure control design of FFN. In this paper, a nonlinear dynamic model of general FFN with fluid resistance, inductance and capacitance is first given. By using this nonlinear model, it is shown that the general FFN dynamics is strictly passive, and the corresponding generalized Hamiltonian form (GHF) is also given. Based on the passivity of FFN, a decentralized adaptive flowrate-pressure coordinated control is newly proposed, taking a simple proportional-integral (PI) form while providing globally asymptotic regulation for network flowrates and pressures. This FFN control method is applied to design the flowrate-pressure controller of an auxiliary startup circuit (ASC) equipped to a modular high temperature gas-cooled reactor (mHTGR), and simulation results in the cases of heating-up and cooling-down show the feasibility of control design.