Browsing by Subject "Networks"

Networked infrastructure-as-a-service testbeds are evolving with higher capacity and more advanced capabilities. Modern testbeds offer stitched virtual circuit capability, programmable dataplanes with software-defined networking (SDN), and in-network processing on adjacent cloud servers. With these capabilities they are able to host virtual network service providers (NSPs) that peer and exchange traffic with edge subnets and with other NSPs in the testbeds. Testbed tenants may configure and program their NSPs to provide a range of functions and capabilities. Programmable NSPs enable innovation in network services and protocols following the pluralist philosophy of network architecture.

Advancing testbeds offer an opportunity to harness their power to deploy production NSPs with topology and value-added features tailored to the needs of specific user communities. For example, one objective of this research is to define abstractions and tools to support built-to-order virtual science networks for data-intensive science collaborations that share and exchange datasets securely at high speed. A virtual science network may combine dedicated high-speed circuits on advanced research fabrics with integrated in-network processing on virtual cloud servers, and links to exchange traffic with customer campus networks and/or testbed slices. We propose security-managed science networks with additional security features including access control, embedded virtual security appliances, and managed connectivity according to customer policy. A security-managed NSP is in essence a virtual software-defined exchange (SDX) that applies customer-specified policy to mediate connectivity.

This dissertation proposes control abstractions for dynamic NSPs, with a focus on managing security in the control plane based on programmable security policy. It defines an architecture for automated NSP controllers that orchestrate and program an NSP's SDN dataplane and manage its interactions with customers and peer NSPs. A key element of the approach is to use declarative trust logic to program the control plane: all control-plane interactions---including route advertisements, address assignments, policy controls, and governance authority---are represented as signed statements in a logic (trust datalog). NSP controllers use a logical inference engine to authorize all interactions and check for policy compliance.

To evaluate these ideas, we develop the ExoPlex controller framework for secure policy-based networking over programmable network infrastructures. An ExoPlex NSP combines a logical NSP controller with an off-the-shelf SDN controller and an existing trust logic platform (SAFE), both of which were enhanced for this project. Experiments with the software on testbeds---ExoGENI, ESnet, and Chameleon---demonstrate the power and potential of the approach. The dissertation presents the research in four parts.

The first part introduces the foundational capabilities of research testbeds that enables the approach, and presents the design of the ExoPlex controller framework to leverage those capabilities for hosted NSPs. We demonstrate a proof-of-concept deployment of an NSP with network function virtualization, an elastic dataplane, and managed traffic security on the ExoGENI testbed.

The second part introduces logical trust to structure control-plane interactions and program security policy. We show how to use declarative trust logic to address the challenges for managing identity, resource access, peering, connectivity and secure routing. We present off-the-shelf SAFE logic templates and rules to demonstrate a virtual SDX that authorizes network stitching and connectivity with logical trust.

The third part applies the controller architecture to secure policy-based interdomain routing among transit NSPs based on a logical trust plane. Signed logic exchanges propagate advertised routes and policies through the network. We show that trust logic rules capture and represent current and evolving Internet security protocols, affording protection equivalent to BGPsec for secure routing and RPKI for origin authentication. The logic also supports programmable policy for managed connectivity with end-to-end trust, allowing customers to permission the NSPs so that customer traffic does not pass through untrusted NSPs (path control).

The last part introduces SCIF, which extends logical peering and routing to incorporate customizable policies to defend against packet spoofing and route leaks. It uses trust logic to define more expressive route advertisements and compliance checks to filter advertisements that propagate outside of their intended scope. For SCIF, we extended the ExoPlex SDN dataplanes to configure ingress packet filters automatically from accepted routes (unicast Reverse Path Forwarding). We present logic templates that capture the defenses of valley-free routing and the Internet MANRS approach based on a central database of route ingress/egress policies (RADb/RPSL). We show how to extend their expressive power for stronger routing security, and complement it with path control policies that constrain the set of trusted NSPs for built-to-order internetworks.

A series of experiments and mathematical models explore the response of the bacteria E. coli to stressors. Experimentally, the effect of L-homocysteine, a non-protein amino acid, is explored, and via math models, the effect of trimethoprim, a common antibiotic, is also explored. Previous work on L-homocysteine labels it a stressor, and this assertion is refined via the presented work. A mathematical model that improves on a previous work published by Kwon et al. (2008) explores the response of E. coli to various supplementations of amino acids when exposed to trimethoprim. New methods of developing antibiotics and therapeutic drug treatments are also explored.

The study of network data in the social and health sciences frequently concentrates on understanding how and why connections form. In particular, the task of determining latent mechanisms driving connection has received a lot of attention across statistics, machine learning, and information theory. In social networks, this mechanism often manifests as community structure. As a result, this work provides methods for discovering and leveraging these communities to better understand networks and the data they generate.

We provide three main contributions. First, we present methodology for performing community detection in challenging regimes. Existing literature has focused on modeling the spectral embedding of a network using Gaussian mixture models (GMMs) in scaling regimes where the ability to detect community memberships improves with the size of the network. However, these regimes are not very realistic. As such, we provide tractable methodology motivated by new theoretical results for networks with non-vanishing noise by using GMMs that incorporate truncation and shrinkage effects.

Further, when covariate information is available, often we want to understand how covariates impact connections. It is likely that the effects of covariates on edge formation differ between communities (e.g. age might play a different role in friendship formation in communities across a city). To address this issue, we introduce a latent space network model where coefficients associated with certain covariates can depend on latent community membership of the nodes. We show that ignoring such structure can lead to either over- or under-estimation of covariate importance to edge formation and propose a Markov Chain Monte Carlo approach for simultaneously learning the latent community structure and the community specific coefficients.

Finally, we consider how community structure can impact experimentation. It is evident that communities can act in different ways, and it is natural that this propagates into experimental design. As as result, this observation motivates our development of community informed experimental design. This design recognizes that information between individuals likely flows along within community edges rather than across community edges. We demonstrate that this design improves estimation of global average treatment effect, even when the community structure of the graph needs to be estimated.

The central question this dissertation engages with is why modern states in the Persian Gulf rely heavily on informal networks of untrained and inexperienced recruits from the region of Balochistan, presently spread across Iran, Afghanistan and Pakistan. The answer, it argues, lies in the longue durée phenomenon of Baloch conquering territories abroad but not ruling in their own name. Baloch, I argue, conquered not to establish their sovereign rule, but to open channels of mobility for others. The rise of nation-states and citizen-armies in the twentieth century limited the possibility of Baloch conquest. Yet, the Baloch continued to find a place in the Gulf’s protection industry through historically shaped informal, familial, commercial, and parapolitical transnational networks. Flexible and persistent Baloch networks provided territorially bounded states the ability to access resources outside their boundaries without investment in formal international contracts.

Moreover, this dissertation makes the argument that mobile Baloch operated as ‘Portfolio-Mercenaries’, offering their military-labor to foreign states in order to build their own portfolio of transnational economic, social and political activities. At times these portfolio projects contradicted state interests; at other moments they corroborated them. In either situation, the non-soldiering activities of mercenaries went on to transform the nature of political order in the twentieth-century space of the Indian Ocean. They shaped the nature of international law, carried state order beyond borders, stabilized unpopular regimes, and provided ready sources of labor. Through the example of Baloch Portfolio-Mercenaries, the dissertation thus highlights the thick and enduring relationships between state and transnational networks.

Critiquing dyads as the unit of analysis in statistical work has become increasingly prominent; a number of scholars have demonstrated that ignoring the interdependencies and selection effects among dyads can bias our inference. My dissertation argues that the problem is even more serious. The bargaining model relies on the assumption that bargaining occurs between two states in isolation. When we relax this assumption one of the most crucial findings of these bargaining models vanishes: it is no longer irrational, even with complete information and an absence of commitment issues, for states to go to war. By accounting for the non-dyadic nature of interstate relations, we are better able to explain a number of empirical realities, and better able to predict when states will go to war.

In the first chapter of my dissertation I model a bargaining episode between three players and demonstrate its marked divergence from canonical bargaining models. In traditional two player bargaining models, it is irrational for states to go to war. I find this irrationality of war to be in part an artifact of limiting the focus to two players. In the model in chapter one, three states are bargaining over policy, and each state has a preference in relation to this policy. When these preferences diverge enough, it can become impossible for players to resolve their disputes peacefully. One implication of this model is that differences between two and three player bargaining is not just a difference in degree, but a difference in kind. The model in this chapter forms the core of the writing sample enclosed. Chapter two tests whether my own model is just an artifact of a particular set of assumptions. I extend the bargaining model to allow for N-players and modify the types of policies being bargained over, and I find that not only do the results hold, in many cases they are strengthened. The second chapter also changes chapter one's model so states are bargaining over resources rather than policy which results in a surprising finding: while we might expect states to be more willing to fight in defense of the homeland than over a policy, if more than two states are involved, it is in fact the disputes over territory that are significantly more peaceful.

In the final chapter of my dissertation, I attempt to apply the insights from the theoretical chapters to the study of interstate conflict and war. In particular, I compare a purely dyadic model of interstate crises to a model that accounts for non-dyadic interdependencies. The non-dyadic model that I present is an Additive and Multiplicative Effects Network model, and it substantially outperforms the traditional dyadic model, both in explaining the variance of the data and in predicting out of sample. By combining the theoretical work in the earlier chapters with the empirical work in the final chapter I can show that not only do dyadic models limit our ability to model the causes of conflict, but that by moving beyond the dyad we actually get notable gains in our ability to understand the world and make predictions.

Bridging the disciplines of art history and cross-cultural studies using a feminist interpretive lens, this dissertation challenges historical narratives of exceptionalism and Eurocentrism through analysis of patterns in sixteenth- and seventeenth-century women’s engagement with expanding worlds through their networks and visual representations of world-traveling people and imported objects. To arrive at an inclusive understanding of the significant relationship between the visual arts and cultural exchange, this dissertation offers a new perspective on the cross-cultural circulation that hallmarked early modern Europe by foregrounding women artists, who as a group have been traditionally excluded from the historical record. A data-driven methodology is used to analyze European women’s access to cross-cultural encounters with cosmopolitan courtiers, enslaved people, and imported curios of Asia, Africa, and the Americas through a database of 249 artists designed and populated for this dissertation. Visual analysis of artworks by a subset of these artists is conducted to understand the relationship between women’s worldly encounters and their subsequent creative choices in depicting artistic subjects that came to be considered exotic, thus positioning the artist as cultural mediator. This research makes a dual contribution: first, it challenges persistent narratives that frame women artists’ access to the world as unavoidably limited by gendered social norms. Second, it constructs a new narrative centered on cross-cultural exchange that moves beyond the limits imposed by traditional accounts that focus exclusively on male artists or treat women artists as anomalies. The first key finding of this dissertation challenges popular narratives that present early modern women artists as “magnificent exceptions” or as products of unusually tolerant environments. This project decenters traditional focal points of individual artists and cities through a database used to map artists’ connections and sites of encounter. The data visualized through maps, graphs, and tables demonstrate the geographic breadth and continuity of European women’s artistic activity. These analyses evince many nodes of activity to support the artists’ cross-cultural exposure. The volume of representations in which women artists engaged with wider worlds demonstrates that they actively participated in the history of cross-cultural circulation, rather than existing outside it. By restoring women’s rightful places in this history, we gain the opportunity to assess whether women artists challenged pre-existing imagery and attitudes of cultural imperialism. The second key finding stems from case studies organized by scales of physical and cultural distance, a structure that enables assessment of the relationship between the intimacy of the artist-subject encounter and the quality of the resulting representation. First, investigation of portraits of world-travelers artists encountered in courtly settings addresses whether women’s depictions aligned with conventional representations of the same subjects. Then, examination of women’s representations of manufactured curiosities and naturalia from Asia, Africa, and the Americas explores how such depictions distinctively relate to European desires for universal power and possession. Finally, this dissertation works to center the erasures of Black figures in visual constructions by women artists in early modern Italy by assessing how women’s representational choices participate in the perpetuation or subversion of pre-existing cultural narratives. These three lines of analysis circumvent the draw to exceptionalize certain figures by focusing on sets of relationships and bringing unnamed figures into the framework. Ultimately, although the artists’ choices conform to some racially biased conventions, they also open the possibility of collaboration with foreign individuals; pay homage to the production of artists from different continents; and create expansive roles for imaginary characters represented as Black. This analysis contributes to our understanding of women’s complex intersectional positions in matrices of variable power and access, and to the debate on their roles as producers of knowledge and culture.

This dissertation addresses two aspects of dynamical systems arising from biological networks: homeostasis-bifurcation and identifiability.

Homeostasis occurs when a biological quantity does not change very much as a parameter is varied over a wide interval. Local bifurcation occurs when the multiplicity or stability of equilibria changes at a point. Both phenomena can occur simultaneously and as the result of a single mechanism. We show that this is the case in the feedback inhibition network motif. In addition we prove that longer feedback inhibition networks are less stable. Towards understanding interactions between homeostasis and bifurcations, we define a new type of singularity, the homeostasis-bifurcation point. Using singularity theory, the behavior of dynamical systems with homeostasis-bifurcation points is characterized. In particular, we show that multiple homeostatic plateaus separated by hysteretic switches and homeostatic limit cycle periods and amplitudes are common when these singularities occur.

Identifiability asks whether it is possible to infer model parameters from measurements. We characterize the structural identifiability properties for feedforward networks with linear reaction rate kinetics. Interestingly, the set of reaction rates corresponding to the edges of the graph are identifiable, but the assignment of rates to edges is not; Permutations of the reaction rates leads to the same measurements. We show how the identifiability results for linear kinetics can be extended to Michaelis-Menten kinetics using asymptotics.

Episodic memory, as a cognitive construct, exists only in relation to those other cognitive constructs that reference it. It is, as Ribot suggests: the tactile, the muscular, the auditory and so forth. And it is even more than this, extending to a breadth of cognitive operations, including, for example, attention and cognitive control, both of which are generally believed to facilitate episodic memory encoding and episodic memory retrieval. Without these types of sensory and cognitive referents, episodic memory does not exist. Accordingly, these types of referents are critical to an understanding of episodic memory. Therefore, in this dissertation I examine how different cognitive constructs serve to facilitate episodic memory.

Chapter 2 examines attention-related subsequent memory effects. Many studies of subsequent memory rely upon a reverse inference, i.e. increased activity in attention-related networks during memory encoding is related to better subsequent memory, ergo increased attention predicts better memory. However, it is only through direct manipulation of attentional states and the examination of specific neural markers that this claim can be strongly established. Additionally, attention is a multifaceted process, and claims that attention in general facilitates memory ignore the fact that attention consists of a set of rapidly enfolding processes. To address these issues, I designed a modified visual-search EEG experiment with a subsequent long-term memory test. The utilization of a visual-search paradigm has advantages, as the search process evokes a series of independent and well-established attention-related EEG markers which can be linked to subsequent memory. All of the attentional effects examined were found to also predict subsequent memory, suggesting that these attentional processes associated with visual search, aid long-term memory formation as well.

Chapter 3 examines how large-scale network dynamics affect long-term memory retrieval. Until now, all studies of long-term memory have focused on individual regions, pair-wise connections between regions, or, very rarely, complex interactions between a small subset of regions. In a pair of fMRI studies, I use mathematical concepts from network science to examine the large-scale brain networks associated with successful remembering and forgetting. In doing so, I demonstrate that the hippocampus increases its integration with the rest brain when individuals successfully remember an item as compared to when they do not.

Chapter 4 examines how individual items are represented in the brain with machine-learning techniques and fMRI data. Studies of episodic memory often focus on things that are common across a set of items, while ignoring the uniqueness of individual events. However, an event’s uniqueness is what defines it as being episodic with respect to memory. A primary reason unique events are not often studied is the difficulty of decoding brain states associated with individual events. In Chapter 4, I develop a machine-learning framework, utilizing cross-subject single-item decoding, to predict what image or word a left-out subject is viewing. This establishes a robust way to detect individual events which could be used in service of better understanding episodic memory.

By examining long-term memory from these perspectives, I provide evidence of how different cognitive constructs facilitate episodic memory. In Chapter 2, I focus on the role of attentional processes with respect to episodic memory encoding, in Chapter 3, I focus on how large-scale network interactions facilitate episodic memory retrieval, and in Chapter 4 I focus on the representational nature of unique events. In all cases, the examination centers on how diverse processes coordinate in order to facilitate episodic memory.

The physics of complex systems has grown considerably as a field in recent decades, largely due to improved computational technology and increased availability of systems level data. One area in which physics is of growing relevance is molecular biology. A new field, systems biology, investigates features of biological systems as a whole, a strategy of particular importance for understanding emergent properties that result from a complex network of interactions. Due to the complicated nature of the systems under study, the physics of complex systems has a significant role to play in elucidating the collective behavior.

In this dissertation, we explore three problems in the physics of complex systems, motivated in part by systems biology. The first of these concerns the applicability of Boolean models as an approximation of continuous systems. Studies of gene regulatory networks have employed both continuous and Boolean models to analyze the system dynamics, and the two have been found produce similar results in the cases analyzed. We ask whether or not Boolean models can generically reproduce the qualitative attractor dynamics of networks of continuously valued elements. Using a combination of analytical techniques and numerical simulations, we find that continuous networks exhibit two effects -- an asymmetry between on and off states, and a decaying memory of events in each element's inputs -- that are absent from synchronously updated Boolean models. We show that in simple loops these effects produce exactly the attractors that one would predict with an analysis of the stability of Boolean attractors, but in slightly more complicated topologies, they can destabilize solutions that are stable in the Boolean approximation, and can stabilize new attractors.

Second, we investigate ensembles of large, random networks. Of particular interest is the transition between ordered and disordered dynamics, which is well characterized in Boolean systems. Networks at the transition point, called critical, exhibit many of the features of regulatory networks, and recent studies suggest that some specific regulatory networks are indeed near-critical. We ask whether certain statistical measures of the ensemble behavior of large continuous networks are reproduced by Boolean models. We find that, in spite of the lack of correspondence between attractors observed in smaller systems, the statistical characterization given by the continuous and Boolean models show close agreement, and the transition between order and disorder known in Boolean systems can occur in continuous systems as well. One effect that is not present in Boolean systems, the failure of information to propagate down chains of elements of arbitrary length, is present in a class of continuous networks. In these systems, a modified Boolean theory that takes into account the collective effect of propagation failure on chains throughout the network gives a good description of the observed behavior. We find that propagation failure pushes the system toward greater order, resulting in a partial or complete suppression of the disordered phase.

Finally, we explore a dynamical process of direct biological relevance: asymmetric cell division in A. thaliana. The long term goal is to develop a model for the process that accurately accounts for both wild type and mutant behavior. To contribute to this endeavor, we use confocal microscopy to image roots in a SHORTROOT inducible mutant. We compute correlation functions between the locations of asymmetrically divided cells, and we construct stochastic models based on a few simple assumptions that accurately predict the non-zero correlations. Our result shows that intracellular processes alone cannot be responsible for the observed divisions, and that an intercell signaling mechanism could account for the measured correlations.