BibTeX
@MISC{Montavista_dynamicpower,
author = {Ibm And Montavista and Montavista Software},
title = {Dynamic Power Management for Embedded Systems},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Implementation The previous Sections presented a high-level design of the DPM architecture. To review, the architecture is a hierarchy of objects: operating points and congruence classes of operating points, operating states, and policies composed of mappings from operating states to congruence classes of operating points. This is a straightforward architecture that could be implemented in an operating system in several ways. This Section gives our preferred implementation and the rationale behind the choices made in the implementation. Although the framework of the power management system described here is simple, accounting for all of the possible interactions of user-level polices and the influence of device constraints is a challenging task. Ultimately, regardless of the implementation, the system designer who creates the power management policies for the system is responsible for understanding all of the constraints imposed by the application with respect to the power management system. Two of the challenges with respect to implementing this system include: . Changes in device constraints may invalidate operating points. Automating these transitions is the primary mechanism by which the architecture relieves the high-level power management task from having to deal with device states. This leads to several obvious conflicts, however.
Keyphrases
operating point dynamic power management embedded system power management system device constraint high-level power management task preferred implementation straightforward architecture dpm architecture previous section system designer possible interaction several way user-level police congruence class high-level design operating system device state challenging task power management policy primary mechanism several obvious conflict
