| 000 | 03140cam a2200349 a 4500 | ||
|---|---|---|---|
| 003 | EG-GiCUC | ||
| 005 | 20250223032920.0 | ||
| 008 | 220213s2021 ua d f m 000 0 eng d | ||
| 040 |
_aEG-GiCUC _beng _cEG-GiCUC |
||
| 041 | 0 | _aeng | |
| 049 | _aDeposite | ||
| 097 | _aM.Sc | ||
| 099 | _aCai01.20.03.M.Sc.2021.Am.T | ||
| 100 | 0 | _aAmira Ahmed Ali Abohozaifa | |
| 245 | 1 | 0 |
_aToward a dynamic internet of things based high performance computing system / _cAmira Ahmed Ali Abohozaifa ; Supervised Abeer Mohamed Elkorany , Ahmed Shawky Moussa |
| 246 | 1 | 5 | _aنحو نظام ديناميكى قائم على إنترنت الأشياء لنظام حوسبة عالية الأداء |
| 260 |
_aCairo : _bAmira Ahmed Ali Abohozaifa , _c2021 |
||
| 300 |
_a92 P. : _bcharts ; _c30cm |
||
| 502 | _aThesis (M.Sc.) - Cairo University - Faculty of Computers and Artificial Intelligence - Department of Computer Sciences | ||
| 520 | _aThe Internet of Things (IoT) became one of the buzzwords in the fields of computer science and technology.The number of connected devices has been growing rapidly which makes it more promising to use these devices for parallel processing. However, the dynamic behaviour of IoT systems, where nodes frequently drop off, may be an obstacle against achieving parallel processing on IoT devices.This thesis discusses the drop off problem, the need for more computational power. Furthermore, it introduces the idea of using up and running IoT systems as HPC infrastructure. In addition to that, it proposes an intelligent redundancy mechanism based on Fuzzy Computing to enhance the IoT system resilience.The introduced mechanism focuses on recovering the lost computation in case of failures inside the system. This mechanism is implemented to achieve resilience to enable using these systems in parallel processing.The recovery of the lost calculations is done by group peering and checkpointing. In the proposed solution, the system nodes are divided into groups. For a given group, the nodes are responsible for recovering the lost calculations of the failed node(s) inside this group. The group size is determined according to the average load of the system which is categorized into 5 fuzzy classes. The mechanism was able to enhance the resilience, achieved up to 96% system resilience even if 75% of the system was lost. In addition to that, the system adaptation communication overhead was considered and enhanced. The fuzzy mechanism provided 16% better performance and 42% less communication than the crisp based method.The communication between nodes was enhanced as well by using fuzzy classes instead of threshold based technique. This technique enhanced the performance by 12% and decreased the number of exchanged message by 61% | ||
| 530 | _aIssued also as CD | ||
| 650 | 0 | _aHPC | |
| 653 | 4 | _aFuzzy Classes | |
| 653 | 4 | _aIoT | |
| 653 | 4 | _aResilience | |
| 700 | 0 |
_aAbeer Mohamed Elkorany , _eSupervisor |
|
| 700 | 0 |
_aAhmed Shawky Moussa , _eSupervisor |
|
| 856 | _uhttp://172.23.153.220/th.pdf | ||
| 905 |
_aNazla _eRevisor |
||
| 905 |
_aShimaa _eCataloger |
||
| 942 |
_2ddc _cTH |
||
| 999 |
_c84201 _d84201 |
||