Thesis (M.S.)--West Virginia University, 2003. / Title from document title page. Document formatted into pages; contains ix, 63 p. : ill. (some col.). Includes abstract. Includes bibliographical references (p. 60-63).

No description available.

published_or_final_version / Electrical and Electronic Engineering / Master / Master of Philosophy

published_or_final_version / Mechanical Engineering / Doctoral / Doctor of Philosophy

Given the popularity of Java, extending the standard Java virtual machine (JVM) to become cluster-aware effectively brings the vision of transparent horizontal scaling of applications to fruition. With a set of cluster-wide JVMs orchestrated as a virtually single system, thread-level parallelism in Java is no longer confined to one multiprocessor. An unmodified multithreaded Java application running on such a Distributed JVM (DJVM) can scale out transparently, tapping into the vast computing power of the cluster. While this notion creates an easy-to-use and powerful parallel programming paradigm, research on DJVMs has remained largely at the proof-of-concept stage where successes were proven using trivial scientific computing workloads only. Real-life Java applications with commercial server workloads have not been well-studied on DJVMs. Their natures including complex and sometimes huge object graphs, irregular access patterns and frequent synchronizations are key scalability hurdles. To design a scalable DJVM for real-life applications, we identify three major unsolved issues calling for a top-to-bottom overhaul of traditional systems. First, we need a more time- and space-efficient cache coherence protocol to support fine-grained object sharing over the distributed shared heap. The recent prevalence of concurrent data structures with heavy use of volatile fields has added complications to the matter. Second, previous generations of DJVMs lack true support for memory-intensive applications. While the network-wide aggregated physical memory can be huge, mutual sharing of huge object graphs like Java collections may cause nodes to eventually run out of local heap space because the cached copies of remote objects, linked by active references, can’t be arbitrarily discarded. Third, thread affinity, which determines the overall communication cost, is vital to the DJVM performance. Data access locality can be improved by collocating highly-correlated threads, via dynamic thread migration. Tracking inter-thread correlations trades profiling costs for reduced object misses. Unfortunately, profiling techniques like active correlation tracking used in page-based DSMs would entail prohibitively high overheads and low accuracy when ported to fine-grained object-based DJVMs. This dissertation presents technical contributions towards all these problems. We use a dual-protocol approach to address the first problem. Synchronized (lock-based) and volatile accesses are handled by a home-based lazy release consistency (HLRC) protocol and a sequential consistency (SC) protocol respectively. The two protocols’ metadata are maintained in a conflict-free, memory-efficient manner. With further techniques like hierarchical passing of lock ownerships, the overall communication overheads of fine-grained distributed object sharing are pruned to a minimal level. For the second problem, we develop a novel uncaching mechanism to safely break a huge active object graph. When a JVM instance runs low on free memory, it initiates an uncaching policy, which eagerly assigns nulls to selected reference fields, thus detaching some older or less useful cached objects from the root set for reclamation. Careful orchestration is made between uncaching, local garbage collection and the coherence protocol to avoid possible data races. Lastly, we devise lightweight sampling-based profiling methods to derive inter-thread correlations, and a profile-guided thread migration policy to boost the system performance. Extensive experiments have demonstrated the effectiveness of all our solutions. / published_or_final_version / Computer Science / Doctoral / Doctor of Philosophy

This thesis makes progress towards the fundamental understanding of heterogeneous and dynamic information systems and the way that we store and process massive data-sets. Reliable large-scale data storage: Distributed storage systems for large clusters typically use replication to provide reliability. Recently, erasure codes have been used to reduce the large storage overhead of three-replicated systems. However, traditional erasure codes are associated with high repair cost that is often considered an unavoidable price to pay. In this thesis, we show how to overcome these limitations. We construct novel families of erasure codes that are optimal under various repair cost metrics, while achieving the best possible reliability. We show how these modern storage codes significantly outperform traditional erasure codes. Low-rank approximations for large-scale data processing: A central goal in data analytics is extracting useful and interpretable information from massive data-sets. A challenge that arises from the distributed and large-scale nature of the data at hand, is having algorithms that are good in theory but can also scale up gracefully to large problem sizes. Using ideas from prior work, we develop a scalable lowrank optimization framework with provable guarantees for problems like the densest k-subgraph (DkS) and sparse PCA. Our experimental findings indicate that this low-rank framework can outperform the state-of-the art, by offering higher quality and more interpretable solutions, and by scaling up to problem inputs with billions of entries. / text

Distributed Feedback (DFB) lasers and integrated laser arrays are of great importance in Wavelength-Division Multiplexing (WDM) systems in fiber optic communication systems. High-performance, low-cost DFB lasers and laser arrays are highly desirable for applications in intra-datacenter transport and in local access networks. This dissertation is focused on the design, fabrication and achievement of high-performance, low-cost DFB Lasers and Integrated DFB Arrays for WDM Systems. It investigates the use of a novel sampled grating approach, called the equivalent phase shift method, to achieve integrated DFB laser arrays with single-mode lasing at uniformly-spaced and precisely-positioned wavelengths. First, laterally-Coupled DFB (LC-DFB) lasers with first-order sidewall gratings are realized, with gratings fabricated by optical interference lithography instead of e-beam. Then, LC-DFB lasers and LC-DFB laser arrays with sampled gratings and equivalent phase shifts are proposed, numerically analyzed and experimentally demonstrated. Each LC-DFB laser with an equivalent quarter-wave phase shift is shown to lase at the pre-specified wavelength in a single longitudinal mode, with good side-mode suppression ratio (SMSR) over a very wide range of injection currents. Integrated LC-DFB laser arrays with five uniformly-spaced wavelength channels are demonstrated, in close agreement with the design. For better performance, buried heterostructure (BH)-DFB laser and laser arrays are also demonstrated using the same sampled-grating technology. A 6-wavelenth laser array with a 300 μm cavity length and a 8-wavelength laser array with 250 μm cavity length are successively demonstrated, each showing precisely positioned lasing wavelengths, good SMSR, and uniformly good lasing characteristics under a wide range of operating currents and temperatures. Finally, it is demonstrated that the wavelength of a monolithic WDM laser array can be continuously tuned over a very wide wavelength range of nearly 40 nm. The proposed method offers a practical and cost-effective solution for the manufacture of high-performance, monolithic multi-wavelength DFB laser arrays as well as widely wavelength-tunable DFB lasers for integrated WDM systems.

ITC/USA 2013 Conference Proceedings / The Forty-Ninth Annual International Telemetering Conference and Technical Exhibition / October 21-24, 2013 / Bally's Hotel & Convention Center, Las Vegas, NV / Distributed Denial of Service (DDoS) attacks have been gaining popularity in recent years. Most research developed to defend against DDoS attacks have focused on analytical studies. However, because of the inherent nature of a DDoS attack and the scale of a network involved in the attack, analytical simulations are not always the best way to study DDoS attacks. Moreover, because DDoS attacks are considered illicit, performing real attacks to study their defense mechanisms is not an alternative. For this reason, using packet/network simulators, such as OPNET Modeler, is the best option for research purposes. Detection of an ongoing DDoS attack, as well as simulation of a defense mechanism against the attack, is beyond the scope of this paper. However, this paper includes design recommendations to simulate an effective defense strategy to mitigate DDoS attacks. Finally, this paper introduces network links failure during simulation in an attempt to demonstrate how the network recovers during and following an attack.

A key debate within leadership research is whether leadership can be conceptualized as a specialized role occupied by individuals or as a shared influence process amongst all members of a group (Yukl, 2006). Since the mid-­‐ 1990s some leadership scholars, as a counterpoint to the dominance of the former and using terms such as shared and distributed leadership, have attempted to elaborate new ‘post-­‐heroic’ leadership models (Badaracco, 2001) of the latter, in which leadership is something that involves all group members. These new forms of leadership are often positioned as something that organizations can implement as part of an adaptive response to a rapidly changing world. Despite a 50-­‐year tradition of construing leadership as a group level construct, little attention has been paid in these emerging debates to the systems psychodynamic perspective. From this perspective there are grounds for suspecting that attempts to implement shared leadership may compound rather than ameliorate issues related to adaptive challenges (Huffington, James and Armstrong, 2004). This thesis engages with the shared and distributed leadership literatures and examines how a systems psychodynamic perspective can contribute not only to debates within these literatures but to the wider controversies in the leadership literature. This thesis reports on the findings of a single, 18-­‐month, longitudinal case study of a senior team whose managing director attempted to implement shared leadership. Using a clinical fieldwork methodology (Schein, 1987) in the systems psychodynamic tradition (Miller, 1993b; Miller and Rice, 1967), this study advances a number of contributions to theory. These include: findings that challenge existing approaches to conceptualizing leadership – shared or otherwise; the elucidation of complex unconscious team processes that are mobilized as a senior team undertakes adaptive work; and thirdly, a more sophisticated and theoretically robust conceptualization of leadership as a group level phenomenon.

Acoustic emissions (AEs) are generated as a result of the creation of, or movement on microcracks in a rock mass. Hypocentres of AEs have been used as a very effective method of visualising the extent (amount and location) of microcrack damage. Studies have used AE locations to investigate both the behaviour of rock failure in laboratory experiments, and to evaluate damage in the Excavation Disturbed Zone (EDZ) around underground openings. The latter has particular significance for the safe storage of nuclear materials in deep underground facilities. Because AEs represent phenomena associated directly with the physical processes occurring in microcracking, then they can also be used to evaluate the fundamental mechanics of the failure. In this thesis a moment tensor (MT) inversion procedure is developed for AEs. This utilises full-waveform records from an array of ultrasonic piezoelectric transducers distributed around the rock mass. The procedure is tested using synthetic amplitudes and is shown to be robust even with high amplitude uncertainties. The inversion is particularly good at resolving the volumetric component in the source. The procedure allows a precise and well-constrained analysis of the forces that are creating the AEs, and, in some cases, that are actually creating the damage. The mechanics can then be related to the stress field in the rock mass, or can be compared to results from dynamic micromechanical models. Three case studies are performed. Two of these investigate the fundamental behaviour of microcracking in the laboratory. A series of laboratory tests are conducted using polyaxial stress to study the mechanics of damage under realistic in situ stress paths. The third case study investigates the mechanics of failure operating in the EDZ. AEs are shown to be truly scaled earthquakes although with often-complex non-double-couple mechanisms.