Loading...
Search
| Friend's email | |
| Your name | |
| Your email | |
| enter code | |
This page was sent successfuly
Digital Object List
Bookmark
Review ListEnergy Management in Energy Harvesting Systems Equipped with Non-Volatile Memories
Hosseinghorban, Ali | 2022
360
Viewed
- Type of Document: Ph.D. Dissertation
- Language: Farsi
- Document No: 55004 (19)
- University: Sharif University of Technology
- Department: Computer Engineering
- Advisor(s): Ejlali, Alireza
- Abstract:
- Forward progress (response time), energy consumption, capacitor size, and correct functionality are the most important design constraints of energy harvesting embedded systems. The system needs to backup the state of the processor and memories during the execution periodically to successfully execute long-running applications on energy harvesting systems with unreliable energy sources and small capacitors. The check-pointing policy, which specifies the place, time, and the data to be backed up, has a major effect on the energy consumption, forward-progress, and correct functionality of energy harvesting systems. Furthermore, emerging non-volatile processors equipped with hybrid volatile and non-volatile memory, improve latency and energy consumption of normal execution and check-pointing in energy harvesting systems. In this dissertation, we discussed the overheads and shortcomings of the state of the art approaches and proposed novel techniques to improve the system. Check-pointing could be static or dynamic where the place and time of check-points are specified in the offline or online phase, respectively. In the static policy, CHANCE manages voltage thresholds of the capacitor, to make a balance between capacitor charging time and failure rate for each task, and it improves the average response time of the system by up to 80%. In the dynamic policy, PROWL and COACH improve the energy overhead of tracking the changes in memory and reduce the number of check-points that are forced to the system to avoid data inconsistency in the memory. COACH changes the architecture of nonvolatile main memory and improves forward-progress by up to 48% without imposing any additional check-point to the system. PROWL modifies the replacement policy of SRAM cache of the system and reduce the number of check-points and response time of the system by up to 57% and 51%, respectively. Furthermore, we proposed the CATNAP-Sim simulator in this dissertation to simulate the behavior of energy harvesting systems.
- Keywords:
- Energy Harvester Systems ; Nonvolatile Memory ; Checkpoint ; Checkpointing with Rollback Recovery ; Energy Management
-
محتواي کتاب
- view