UMD Theses and Dissertations
Permanent URI for this collectionhttp://hdl.handle.net/1903/3
New submissions to the thesis/dissertation collections are added automatically as they are received from the Graduate School. Currently, the Graduate School deposits all theses and dissertations from a given semester after the official graduation date. This means that there may be up to a 4 month delay in the appearance of a given thesis/dissertation in DRUM.
More information is available at Theses and Dissertations at University of Maryland Libraries.
Browse
6 results
Search Results
Item Distributed Load Balancing Algorithm in Wireless Networks(2014) Sheikhattar, Alireza; Kalantari, Mehdi; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)As communication networks scale up in size, complexity and demand, effective distribution of the traffic load throughout the network is a matter of great importance. Load balancing will enhance the network throughput and enables us to utilize both communication and energy resources more evenly through an efficient redistribution of traffic load across the network. This thesis provides an algorithm for balancing the traffic load in a general network setting. Unlike most of state-of-the-art algorithms in load balancing context, the proposed method is fully distributed, eliminating the need to collect information at a central node and thereby improving network reliability. The effective distribution of load is realized through solving a convex optimization problem where the p-norm of network load is minimized subject to network physical constraints. The optimization solution relies on the Alternating Direction Method of Multipliers (ADMM), which is a powerful tool for solving distributed convex optimization problems. A three-step ADMM-based iterative scheme is derived from suitably reformulated form of p-norm problem. The distributed implementation of the proposed algorithm is further elaborated by introducing a projection step and an initialization setup. The projection step involves an inner-loop iterative scheme to solve linear subproblems. In a distributed setting, each iteration step requires communication among all neighboring nodes. Due to high energy consumption of node-to-node communication, it is most appealing to devise a fast and computationally efficient iterative scheme which can converge to optimal solution within a desired accuracy by using as few iteration steps as possible. A fast convergence iterative scheme is presented which shows superior convergence performance compared to conventional methods. Inspired by fast propagation of waves in physical media, this iterative scheme is derived from partial differential equations for propagation of electrical voltages and currents in a transmission line. To perform these iterations, all nodes should have access to an acceleration parameter which relies on the network topology. The initialization stage is developed in order to overcome the last challenging obstacle toward achieving a fully distributed algorithm.Item ENABLING GEOGRAPHICALLY DISTRIBUTED, INTERGENERATIONAL, CO-OPERATIVE DESIGN(2012) Walsh, Gregory; Druin, Allison; Library & Information Services; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)As more children's technologies are designed to be used with a global audience, new technologies need to be created to include more children's voices in the design process. However, working with those who that are geographically distributed as design partners is difficult because existing technologies do not support this process, do not enable distributed design, or are not child-friendly. In this dissertation, I take a research-through-design approach to develop an online environment that enables geographically distributed, intergenerational co-operative design. I began my research with participant-observations of in-person, co-located intergeneration co-operative design sessions that used Cooperative Inquiry techniques at the University of Maryland. I then analyzed those observations, determined a framework that occurs during in-person design sessions and developed a prototype online design environment based on that scaffolding. With the initial prototype deployed to a geographic distributed, intergenerational co-design team, I employed Cooperative Inquiry to design new children's technologies with children. I iteratively developed the prototype environment over eight weeks to better support geographically distributed co-design. Adults and children participated in these design sessions and there was no significant difference between the children and adults in the number of design sessions in which they chose to participate. After the design research on the prototype was complete, I interviewed the child participants who were in the online intergenerational design team to better understand their experiences. During the interviews, I found that the child participants had strong expectations of social interaction within the online design environment and were frustrated by the lack of seeing other participants online at the same time. In order to alleviate this problem, five of the participants involved their families in some way in the design process and created small, remote intergenerational design teams to compensate for the perceived shortcomings of the online environment. I compared Online Kidsteam with in-person Kidsteam to evaluate if the online environment was successful in supporting geographically-distributed, intergeneration co-design. I found that although it was not the same in terms of the social aspects of in-person Kidsteam, it was successful in its ability to include more people in the design process.Item GENERALIZED DISTRIBUTED CONSENSUS-BASED ALGORITHMS FOR UNCERTAIN SYSTEMS AND NETWORKS(2010) Matei, Ion; Baras, John S; Martins, Nuno C; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)We address four problems related to multi-agent optimization, filtering and agreement. First, we investigate collaborative optimization of an objective function expressed as a sum of local convex functions, when the agents make decisions in a distributed manner using local information, while the communication topology used to exchange messages and information is modeled by a graph-valued random process, assumed independent and identically distributed. Specifically, we study the performance of the consensusbased multi-agent distributed subgradient method and show how it depends on the probability distribution of the random graph. For the case of a constant stepsize, we first give an upper bound on the difference between the objective function, evaluated at the agents' estimates of the optimal decision vector, and the optimal value. In addition, for a particular class of convex functions, we give an upper bound on the distances between the agents' estimates of the optimal decision vector and the minimizer and we provide the rate of convergence to zero of the time varying component of the aforementioned upper bound. The addressed metrics are evaluated via their expected values. As an application, we show how the distributed optimization algorithm can be used to perform collaborative system identification and provide numerical experiments under the randomized and broadcast gossip protocols. Second, we generalize the asymptotic consensus problem to convex metric spaces. Under minimal connectivity assumptions, we show that if at each iteration an agent updates its state by choosing a point from a particular subset of the generalized convex hull generated by the agents current state and the states of its neighbors, then agreement is achieved asymptotically. In addition, we give bounds on the distance between the consensus point(s) and the initial values of the agents. As an application example, we introduce a probabilistic algorithm for reaching consensus of opinion and show that it in fact fits our general framework. Third, we discuss the linear asymptotic consensus problem for a network of dynamic agents whose communication network is modeled by a randomly switching graph. The switching is determined by a finite state, Markov process, each topology corresponding to a state of the process. We address both the cases where the dynamics of the agents are expressed in continuous and discrete time. We show that, if the consensus matrices are doubly stochastic, average consensus is achieved in the mean square and almost sure senses if and only if the graph resulting from the union of graphs corresponding to the states of the Markov process is strongly connected. Fourth, we address the consensus-based distributed linear filtering problem, where a discrete time, linear stochastic process is observed by a network of sensors. We assume that the consensus weights are known and we first provide sufficient conditions under which the stochastic process is detectable, i.e. for a specific choice of consensus weights there exists a set of filtering gains such that the dynamics of the estimation errors (without noise) are asymptotically stable. Next, we develop a distributed, sub-optimal filtering scheme based on minimizing an upper bound on a quadratic filtering cost. In the stationary case, we provide sufficient conditions under which this scheme converges; conditions expressed in terms of the convergence properties of a set of coupled Riccati equations. We continue by presenting a connection between the consensus-based distributed linear filter and the optimal linear filter of a Markovian jump linear system, appropriately defined. More specifically, we show that if the Markovian jump linear system is (mean square) detectable, then the stochastic process is detectable under the consensus-based distributed linear filtering scheme. We also show that the optimal gains of a linear filter for estimating the state of a Markovian jump linear system, appropriately defined, can be used to approximate the optimal gains of the consensus-based linear filter.Item Using MODIS Satellite Images to Confirm Distributed Snowmelt Model Results in a Small Arctic Watershed(2009) Choy, David F.; Brubaker, Kaye; Civil Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Environmental analysts face the problem of obtaining distributed measurements to evaluate the performance of models with increasingly small spatiotemporal resolution. While U.S. government agencies readily provide both measurement products and data tools for the study of global change occurring over entire seasons and across continental areas, analysts need access to the low-level data that provides the basis for global products. Finally, analysts need to consider sensor errors inherent in low-level products that are accounted for in global, composite products. Hydrologists using tools for managing low-level snow swath measurements, in particular, must consider how measurements are affected by sensor errors like snow-cloud confusion and sensor errors due to low ground illumination at night. This thesis aims to explore the use of remotely sensed snow maps to confirm a time series of model maps. Specifically, snow covered area (SCA) measurements remotely sensed by the National Aeronautics and Space Administration (NASA) are used to confirm SCA predictions modeled by the United States Agriculture Department (USDA). The measurements come from the two Moderate Resolution Imaging Spectroradiometer (MODIS) sensors aboard near-polar, sun-synchronous satellites named Aqua and Terra. The USDA calls the model TOPMODEL-Based Land-Atmosphere Transfer Scheme (TOPLATS). The Upper Kuparuk River Watershed (UKRW) on the North Slope of Alaska acts as the case study location. To meet the map-comparison goal, the Kappa statistic, Kappa statistic variants, and probability density functions expressing measurement uncertainty in discrete scenes all evaluate the ability of MODIS measurements to confirm the accuracy of TOPLATS model maps. Data management objectives to make measured data accessible and comparable to the model output comprise a supporting goal. Results show that individual composite statistics, like the proportion of agreement between two maps, can easily obscure spatiotemporally distributed confirmation information without additional statistics and side-by-side images of measurement maps and model maps. These tools show some promise for using MODIS to confirm model predictions of snowmelt that occur across less than 150 km2 and less than a few days, however, clouds and malfunctioning sensors limit such use.Item The Lattice Project: A Multi-model Grid Computing System(2009) Bazinet, Adam Lee; Cummings, Michael P; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)This thesis presents The Lattice Project, a system that combines multiple models of Grid computing. Grid computing is a paradigm for leveraging multiple distributed computational resources to solve fundamental scientific problems that require large amounts of computation. The system combines the traditional Service model of Grid computing with the Desktop model of Grid computing, and is thus capable of utilizing diverse resources such as institutional desktop computers, dedicated computing clusters, and machines volunteered by the general public to advance science. The production Grid system includes a fully-featured user interface, support for a large number of popular scientific applications, a robust Grid-level scheduler, and novel enhancements such as a Grid-wide file caching scheme. A substantial amount of scientific research has already been completed using The Lattice Project.Item Flexible and Efficient Control of Data Transfers for Loosely Coupled Components(2008-04-26) Wu, Shang-Chieh; Sussman, Alan; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Allowing loose coupling between the components of complex applications has many advantages, such as flexibility in the components that can participate and making it easier to model multiscale physical phenomena. To support coupling of parallel and sequential application components, I have designed and implemented a loosely coupled framework which has the following characteristics: (1) connections between participating components are separately identified from the individual components, (2) all data transfers between data exporting and importing components are determined by a runtime-based low overhead method (approximate match), (3) two runtime-based optimization approaches, collective buffering and inverse-match cache, are applied to speed up the applications in many common coupling modes, and (4) a multi-threaded multi-process control protocol that can be systematically constructed by the composition of sub-tasks protocols. The proposed framework has been applied to two real world applications, and the deployment approach and runtime performance are also studied. Currently the framework runs on x86 Linux clusters, and porting strategies for multicore x86 processors and advanced high performance computer architectures are also explored.