USMA Digital Commons (United States Military Academy, West Point)
Not a member yet
1355 research outputs found
Sort by
Numerical Analysis of a Combustion Model for Layered Media Via Mathematical Homogenization
Get PDFWe propose to investigate a mathematical modelfor combustion in a rod made of periodically alternating thinlayers of two combustible materials such as those occurring ingun propellants. We apply the homogenization theory to resolvethe fast oscillations of the model’s coefficients across adjacentlayers, and set up numerical simulations to better understandthe reactions occurring in such media
Modeling Vegetation-Erosion Dynamics using Differential Equations with Human Factors
Get PDFThe effects of soil erosion are often devastating. Plants can reduce erosion by slowing runoff and reinforcing soil using its roots. In this project, we investigate the dynamic relationship between vegetation and erosion processes. We assume an inverse relationship between vegetation density and soil erosion: that is, an increase in vegetation cover reverses soil degradation and a decrease in vegetation cover intensifies the problem of erosion. We also assume that human activities (like logging, road-building) affect both vegetation development and resilience against erosion. Our model for the vegetation-erosion dynamics is a two-dimensional nonlinear system of differential equations with logistic growth on both variables. Equilibrium and nullcline analysis methods are applied to determine all possible dynamic scenarios between vegetation and erosion. The resulting parameter conditions can be used to analyze bifurcations on the vegetation and erosion dynamics
Cyber Creative Generative Adversarial Network for Novel Malicious Packets
Get PDFMachine learning (ML) requires both quantity and variety of examples in order to learn generalizable patterns. In cybersecurity, labeling network packets is a tedious and difficult task. This leads to insufficient labeled datasets of network packets for training ML-based Network Intrusion Detection Systems (NIDS) to detect malicious intrusions. Furthermore, benign network traffic and malicious cyber attacks are always evolving and changing, meaning that the existing datasets quickly become obsolete. We investigate generative ML modeling for network packet synthetic data generation/augmentation to improve NIDS detection of novel, but similar, cyber attacks by generating well-labeled synthetic network traffic. We develop a Cyber Creative Generative Adversarial Network (CCGAN), inspired by previous generative modeling to create new art styles from existing art images, trained on existing NIDS datasets in order to generate new synthetic network packets. The goal is to create network packet payloads that appear malicious but from different distributions than the original cyber attack classes. We use these new synthetic malicious payloads to augment the training of a ML-based NIDS to evaluate whether it is better at correctly identifying whole classes of real malicious packet payloads that were held-out during classifier training. Results show that data augmentation from CCGAN can increase a NIDS baseline accuracy on a novel malicious class from 79% to 97% with a minimal degradation in accuracy on benign classes (98.9% to 98.7%)
Military and Security Applications: Cybersecurity (Encyclopedia of Optimization, Third Edition)
Get PDFThe domain of cybersecurity is growing as part of broader military and security applications, and the capabilities and processes in this realm have qualities and characteristics that warrant using solution methods in mathematical optimization. Problems of interest may involve continuous or discrete variables, a convex or non-convex decision space, differing levels of uncertainty, and constrained or unconstrained frameworks. Cyberattacks, for example, can be modeled using hierarchical threat structures and may involve decision strategies from both an organization or individual and the adversary. Network traffic flow, intrusion detection and prevention systems, interconnected human-machine interfaces, and automated systems – these all require higher levels of complexity in mathematical optimization modeling and analysis. Attributes such as cyber resiliency, network adaptability, security capability, and information technology flexibility – these require the measurement of multiple characteristics, many of which may involve both quantitative and qualitative interpretations. And for nearly every organization that is invested in some cybersecurity practice, decisions must be made that involve the competing objectives of cost, risk, and performance. As such, mathematical optimization has been widely used and accepted to model important and complex decision problems, providing analytical evidence for helping drive decision outcomes in cybersecurity applications. In the paragraphs that follow, this chapter highlights some of the recent mathematical optimization research in the body of knowledge applied to the cybersecurity space. The subsequent literature discussed fits within a broader cybersecurity domain taxonomy considering the categories of analyze, collect and operate, investigate, operate and maintain, oversee and govern, protect and defend, and securely provision. Further, the paragraphs are structured around generalized mathematical optimization categories to provide a lens to summarize the existing literature, including uncertainty (stochastic programming, robust optimization, etc.), discrete (integer programming, multiobjective, etc.), continuous-unconstrained (nonlinear least squares, etc.), continuous-constrained (global optimization, etc.), and continuous-constrained (nonlinear programming, network optimization, linear programming, etc.). At the conclusion of this chapter, research implications and extensions are offered to the reader that desires to pursue further mathematical optimization research for cybersecurity within a broader military and security applications context
Effects of location, classroom orientation, and air change rate on potential aerosol exposure: an experimental and computational study
No full textThis study examined the dispersion of potentially infectious aerosols in classrooms by means of both a CO2 tracer gas, and multizone contaminant transport modeling. A total of 20 tests were conducted in three different university classrooms at multiple air change rates (4.4-9.7/hr), each with two different room orientations: one with the tracer gas released from six student desks toward the air return, and one with the same tracer gas released away from it. Resulting tracer concentrations were measured by 19 different monitors arrayed throughout the room. Steady-state, mean tracer gas concentrations were calculated in six instructor zones (A-F) around the periphery of the room, with the results normalized by the concentration at the return, which was assumed to be representative of the well-mixed volume of the room. Across all classrooms, zones farthest from the return (C,D) had the lowest mean normalized concentrations (0.75), while those closest to the return (A,F) had the highest (0.95). This effect was consistent across room orientations (release both toward and away from the return), and air change rates. In addition, all zones around the periphery of the room had a significantly lower concentration than those adjacent to the sources. Increasing the ventilation rate reduced tracer gas concentrations significantly. Similar trends were observed via a novel approach to CONTAM modeling of the same rooms. These results indicate that informed selection of teaching location within the classroom could reduce instructor exposure.
Environmental significance
Maintaining healthy air in classrooms is critical to preventing the spread of airborne viruses, and instructors are often in higher risk categories than students. This study indicates that higher ventilation rates coupled with careful selection of teaching location (on the periphery of the room, farthest from the return if possible) could potentially decrease instructor exposure to infectious aerosols
Geospatial Big Data Analytics for Quality Control of Surveys
No full textGeospatial big data analytics allows survey quality control analysts to draw important conclusions about survey data quality that otherwise would take excessive time and resources. In this work, we explored two algorithmic methods that can help ensure reliability of survey interviews by detecting geospatial outliers. Focusing on geospatial data collected from surveys, we implemented outlier detection techniques with two different distance metrics to identify statistical anomalies in real-world datasets that may have qualitative interpretations. We found that one algorithm, which considers the local distribution of points in a dataset, identifies a different set of outliers when compared to another method, which considers the global distribution of points. Since there was a small overlap (10-19%) of flagged points between the two algorithms implemented, it may be helpful for analysts to focus on the fewer “outlier” points that are flagged by both methods rather than all the “outlier” points that are flagged by each algorithm. Finally, analysts should consider the computational costs, as the algorithms differ significantly
A Potemkin Military? Russia’s Over-Estimated Legions
No full textSeen from the West, the Red Army was an able, well-integrated, and competent opponent able to rapidly launch joint offensive operations with no or little warning. Every Western soldier learned how the Soviets would fight by watching Red Army propaganda movies which projected a fast-moving armored onslaught that would either overrun any defense or destroy the defending forces after encirclement. The West was taken in by the Communist propaganda machine because these were the only movies showing Soviet capabilities. We believed that Russianarmored divisions would sweep across the open landscape, cross rivers and streams with ease when engineers unfolded pontoon bridges as the spearhead arrived, all surrounded by a symphony of well-orchestrated artillery and rocket fire and framed by the smoke trails of SU-24 ground attack aircraft in joint operations. We looked at the Soviet, and now Russian, order of battle and drew mathematical inference – and ended up being wrong. We missed discipline, leadership, coordination, trust, and the effects on troops and hardware, living in a culture of corruption and theft for decades
Goodbye Vladivostok, Hello Hǎishēnwǎi!
No full textChinese leaders may well have a better understanding of Russian decline than their Western counterparts. Russia is not a superpower but is treated as a superpower. There is a reason why Russia constantly threatens the use of nuclear arms; the country is an economic dwarf compared to its geographical size. Russian GDP is only nine times greater than the revenues of the American retailer Costco. The economic strength of Russia is limited; natural resource sales produce cash flow that can be distributed to keep the Russian people calm and elites happy, but not much else. Almost no one outside Russia seeks to buy Russian cars, household appliances, or TVs. It confounds logic for China to treat Russia as a superpower and an equal when Russia has a tenth of the economic torque unless China itself has fallen for the Russian propaganda and sable rattling. Sun Yat-Sen, the nationalist father of modern China, pushed for the decolonization of the Russian Far East and the resettlement of Han and Manchu Chinese on their ancestral lands until the Chinese Revolution halted further discussion
Generating Realistic Cyber Data for Training and Evaluating Machine Learning Classifiers for Network Intrusion Detection Systems
No full textCyberspace operations, in conjunction with artificial intelligence and machine learning enhanced cyberspace infrastructure, make it possible to connect sensors directly to shooters independent of human control. These technologies serve as the pivot around which cyber data from the military’s Internet of Battlefield Things, for example, will be turned into actionable insight and knowledge and, ultimately, an information advantage for the military. As such, network intrusion detection systems must detect, evaluate, and respond to malicious cyber traffic at machine speed. Generative adversarial networks and variational autoencoders are fit as generative models with labeled cyber data from a real military enterprise network. These generative models are used to create realistic, synthetic cyber data. A combination of real and synthetic cyber data sets are then used to train several machine learning models for network intrusion detection. Purely synthetic data is shown to be statistically similar to the real data. There is no statistically significant difference in the performance of classifiers trained with real data versus a combination of real and synthetic data; however, classifiers trained with only synthetic data underperformed. To avoid a decrease in intrusion detection performance, classifiers must be trained with at least 15% real data