Theses and Dissertations from UMD
Permanent URI for this communityhttp://hdl.handle.net/1903/2
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 give thesis/dissertation in DRUM
More information is available at Theses and Dissertations at University of Maryland Libraries.
Browse
3 results
Search Results
Item Online Network Design under Uncertainty(2017) Dehghani, Sina; Hajiaghayi, MohammadTaghi; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Today, computer and information networks play a significant role in the success of businesses, both large and small. Networks provide access to various services and resources to end users and devices. There has been extensive research on de- signing networks according to numerous criteria such as cost-efficiency, availability, adaptivity, survivability, among others. In this dissertation, we revisit some of the most fundamental network design problems in the presence of uncertainty. In most realistic models, we are forced to make decisions in the presence of an incomplete input, which is the source of uncertainty for an optimization algorithm. There are different types of uncertainty. For example, in stochastic settings, we may have some random variables derived from some known/unknown distributions. In online settings, the complete input is not known in a-priori and pieces of the input become available sequentially; leaving the algorithm to make decisions only with partial data. In this dissertation, we consider network design and network optimization problems with uncertainty. In particular, we study online bounded-degree Steiner network design, online survivable network design, and stochastic k-server. We analyze their complexity and design competitive algorithms for them.Item Dual-Based Local Search for Deterministic, Stochastic and Robust Variants of the Connected Facility Location Problem(2011) Bardossy, Maria G.; Raghavan, Subramanian; Business and Management: Decision & Information Technologies; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)In this dissertation, we propose the study of a family of network design problems that arise in a wide range of practical settings ranging from telecommunications to data management. We investigate the use of heuristic search procedures coupled with lower bounding mechanisms to obtain high quality solutions for deterministic, stochastic and robust variants of these problems. We extend the use of well-known methods such as the sample average approximation for stochastic optimization and the Bertsimas and Sim approach for robust optimization with heuristics and lower bounding mechanisms. This is particular important for NP-complete problems where even deterministic and small instances are difficult to solve to optimality. Our extensions provide a novel way of applying these techniques while using heuristics; which from a practical perspective increases their usefulness.Item Optimization of Contemporary Telecommunications Networks: Generalized Spanning Trees and WDM Optical Networks(2005-12-05) Stanojevic, Daliborka; Raghavan, Subramanian; Decision and Information Technologies; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)We present a study of two NP-hard telecommunications network design problems - the prize-collecting generalized minimum spanning tree problem (PCGMST) and the design of optical networks with wavelength division multiplexing. The first problem, the PCGMST problem, involves the design of regional backbone networks, where a set of local area networks (LANs) need to be connected by a minimum cost tree network using exactly one gateway site from each LAN. We present several polynomial time heuristics for the PCGMST problem and show that these algorithms, at best, provide only modest quality solutions. We also present two metaheuristics - a local search procedure and a genetic algorithm, and show that these procedures provide compelling high-quality results on a large set of test problems. Our study of the PCGMST problem is concluded by a presentation of two exact solution procedures that can be used to find optimal solutions in networks of moderate size. The second problem studied in this dissertation is a more complex network design problem that involves optical networks with wavelength division multiplexing (WDM). These networks provide an abundance of transmission bandwidth, but require the use of expensive equipment, which, in turn, mandates careful use of the resources available for their design. The novel aspect of WDM optical networks is that they require simultaneous design of two network layers. The first layer is the virtual topology that requires routing of logical paths over the physical layer of optical fibers. The second layer involves routing and grooming of traffic requests over the logical paths established in the virtual topology. This problem has been extensively studied in the last 10 years, but due to its notoriously hard nature, only few exact solution procedures for relaxed versions of this problem were developed so far. We propose one exact and two approximate branch-and-price algorithms for two versions of the WDM optical network design problem and present results of the computational study involving two different design objectives. Finally, we propose two classes of valid inequalities for our branch-and-price algorithms, and discuss applicability of our algorithms to different versions of the WDM optical network design problem.