| dc.identifier.uri | http://hdl.handle.net/11401/77491 | |
| dc.description.sponsorship | This work is sponsored by the Stony Brook University Graduate School in compliance with the requirements for completion of degree. | en_US |
| dc.format | Monograph | |
| dc.format.medium | Electronic Resource | en_US |
| dc.language.iso | en_US | |
| dc.publisher | The Graduate School, Stony Brook University: Stony Brook, NY. | |
| dc.type | Dissertation | |
| dcterms.abstract | Grid systems are widely used to transfer power and information in various forms in many engineering and scientific areas such as grid computing systems, electrical grids, control grid and etc. A good handling of task partition, task allocation and load balancing can significantly increase a grid systems' efficiency. In this dissertation, balancing the loads in electrical grid systems and optimizing grid computing systems are analyzed. Unbalanced loads on feeders increase power system investment and operating costs. Three-phase lateral loads phase swapping is one of the popular methods to balance such systems. We employed a dynamic programming algorithm that makes optimal suggestions to balance the load in electrical grid systems given an input of previous years' data. The algorithm is compared with exhaustive search, the greedy algorithm and heuristic algorithms and it excels in terms of optimality and running time. Based on this, a more general load balancing algorithm with spatial consideration for electrical grid is developed. For the grid computing systems, an interesting class of research topics is the optimal task partition and their mapping to different distributed computing machines with communication time that is nonlinear to the size of the transferring files. Grid computing systems are essentially distributed computing systems without workload dependencies on different machines and with internal communications. Thus, Divisible Load Theory (DLT) is a good match to the scheduling problems in grid computing systems. We developed a DLT-based method to optimally partition the computing load into fractions and map them to computing machines with nonlinear communication speed in the size of loads. Furthermore, two novel performance measurements for grid computing systems with multi-level tree networks are examined. One measure is utilization: the fraction of time processors are busy processing computational load. The other is progress: the percentage of load processed so far at a given time. A variety of scheduling policies are considered. | |
| dcterms.available | 2017-09-20T16:52:48Z | |
| dcterms.contributor | Tang, Wendy | en_US |
| dcterms.contributor | Robertazzi, Thomas G. | en_US |
| dcterms.contributor | Gamboa, Calos | en_US |
| dcterms.contributor | Arkin, Ether. | en_US |
| dcterms.creator | Wang, Kai | |
| dcterms.dateAccepted | 2017-09-20T16:52:48Z | |
| dcterms.dateSubmitted | 2017-09-20T16:52:48Z | |
| dcterms.description | Department of Electrical Engineering. | en_US |
| dcterms.extent | 143 pg. | en_US |
| dcterms.format | Monograph | |
| dcterms.format | Application/PDF | en_US |
| dcterms.identifier | http://hdl.handle.net/11401/77491 | |
| dcterms.issued | 2013-12-01 | |
| dcterms.language | en_US | |
| dcterms.provenance | Made available in DSpace on 2017-09-20T16:52:48Z (GMT). No. of bitstreams: 1
Wang_grad.sunysb_0771E_11632.pdf: 2109994 bytes, checksum: 423eeeaea3a009596bf559cb46fc22c2 (MD5)
Previous issue date: 1 | en |
| dcterms.publisher | The Graduate School, Stony Brook University: Stony Brook, NY. | |
| dcterms.subject | Grid Computing, Load balancing, Optimization, Scheduling, Smart Grid | |
| dcterms.subject | Electrical engineering | |
| dcterms.title | On the optimization of Grid Systems | |
| dcterms.type | Dissertation | |
