http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
- 中文 을 입력하시려면 zhongwen을 입력하시고 space를누르시면됩니다.
- 北京 을 입력하시려면 beijing을 입력하시고 space를 누르시면 됩니다.
RISS 인기검색어
- 검색키워드
- 저자 : Nine, Md S. Q. Zulkar (검색결과 3 건)
-
-
Nine Birk, Cara The University of Arizona 2006 해외박사(DDOD)
Traditional theories of domestic distributive justice take two claims for granted. (1) State territorial borders place legitimate limits on the scope of obligations of distributive justice, i.e., there is an obligation to distribute goods within our territory but not beyond our territory. (2) States have a need for and a legitimate claim to exclusive territorial jurisdiction. Given increasing globalization and the recent prominence of international theories of distributive justice, it is now obvious that these two claims cannot be taken for granted. Theories of distributive justice must explain how and why state borders affect distributive obligations. In this dissertation I argue that state borders serve fundamental values in a liberal theory of justice. As such, state borders are morally relevant to a theory of justice. I argue for a Lockean theory of territory; state territory is justified because it serves four fundamental Lockean values of need, efficiency, the labor theory of desert, and self-determination. State borders mark the boundaries of a state's autonomous territory. State territory, and the borders that mark the boundaries of that territory, are valuable in a liberal theory of justice. This conclusion has implications for the answer to the question: what is owed to foreigners? The fundamental values served by the state's right to territory also support the state's right to control the natural resources within its territory and the state's right to control benefits that flow from the resources within the territory. This means that the state has a right to distribute the benefits from the resources within its territory and (to some degree) to exclude foreigners from these benefits.
-
Multi-objective Optimization of Big Data Transfers
Nine, Md S. Q. Zulkar State University of New York at Buffalo ProQuest D 2020 해외박사(DDOD)
The amount of data moved over dedicated and non-dedicated network links has been increasing at a much faster rate than the increase in the network capacity. However, the existing solutions fail to guarantee even the promised achievable transfer throughput. High throughput data transfers require increased utilization of the underlying resources with side effects such as high-energy consumption at the end systems and network infrastructure. As a result, the global data movement over the Internet generates an energy footprint of 100 terawatt-hours per year, costing billions of dollars to the world economy. Although a considerable amount of research has rendered power management techniques for the hardware-level networking infrastructure, there has not been much prior work focusing on the joint optimization of data transfer throughput and energy consumption at the end systems.In this dissertation, we propose a novel approach for multi-objective optimization of big-data transfers and high-speed packet processing. In our work, we consider the following optimization objectives: (1) data transfer throughput, (2) energy efficiency, (3) Service Layer Agreement (SLA) based joint optimization of achievable throughput and energy consumption, (4) device-specific optimization, (5) cross-layer throughput and energy optimization, and (5) energy-efficient optimization of high-speed packet processing. To achieve these goals, we investigate various factors that can affect the data transfer performance and energy consumption in both compute servers and mobile devices, such as – concurrency, parallelism, pipelining, buffer size, CPU core allocation, dynamic frequency scaling, last level cache (LLC) size, and DMA buffer size. We have collected real data transfer logs to analyze the impact of various factors on the data transfer task and high-speed packet processing.Our novel approach to end-system optimization for data transfer and energy efficiency is based on mathematical modeling with offline knowledge discovery and adaptive online decision making. In the offline analysis phase, we mine the historical data transfer logs to gather knowledge about the data transfer characteristics. The online adaptation phase uses the discovered knowledge from the offline analysis, along with the real-time feedback of the network condition, to dynamically optimize the protocol parameters. As the real-time investigation is expensive and provides only partial knowledge about the current network status, our model uses historical knowledge and data about the network to reduce the real-time investigation overhead while ensuring near-optimal throughput for each transfer. We also explored various challenges and optimization opportunities in mobile data transfers, where energy efficiency is crucial. Then we investigated cross-layer optimization of data transfer where we jointly optimized application layer parameters with lower-layer parameters. We also studied extensively high-speed packet processing framework (e.g., Network Function Virtualization) and proposed Reinforcement Learning (RL) based framework to optimize its throughput and energy efficiency.Our novel approach on throughput optimization is tested over different networks with different datasets, and it has outperformed its closest competitor by 1.7x and the default case by 5x. It also achieved up to 93% accuracy compared to the optimal achievable throughput possible on those networks. Our energy-efficient solution for bulk data transfers, GreenDataFlow, supports SLAs that give the service providers and the consumers the ability to fine-tune their goals and priorities in this optimization process. Our preliminary results show that GreenDataFlow outperforms the closest-competing state-of-the-art solution in this area by 50% for energy-saving and by 2.5x for the achieved end-to-end transfer performance. Moreover, our mobile device-specific solution, FastHLA, can achieve significant energy savings at the application layer during mobile network I/O, without sacrificing the performance. FastHLA can increase data transfer throughput by up to 10x and decrease energy consumption by up to 5x compared to state-of-the-art solutions. Our cross-layer optimization algorithms outperform state-of-the-art solutions, achieving up to 80% higher throughput while consuming 48% less energy. Our energy-efficient packet processing model, GreenNFV, achieves 4x throughput improvement over the baseline settings and 1.5x energy efficiency.
연관 검색어 추천
이 검색어로 많이 본 자료
활용도 높은 자료
