Periodic power management schemes for real-time event streams

Periodic thermal management for hard real-time systems

“…I [11], [19]. The task streams set studied in [18], [22] is used for our case studies and the parameters are summarized in Tab. II.…”

“…According to the hardware model, we directly know that t o f f has to be no less than t swo f f to cover the timing overhead of model-switching. To avoid situations similar to the example, t o f f must be bounded by an upper bound, which is calculated according to the maximum service curve in [18]:…”

“…Thus, we transform the event-based arrival curveᾱ(∆) to time-based arrival curve α(∆), which describes the amount of execution time demanded by arrival events. Suppose that the worst-case execution time of one event in arrival stream is c, then the arrival curve transformation can be performed as α u (∆) = c ·ᾱ u (∆) and α l (∆) = c ·ᾱ l (∆) [18].…”

“…To reduce the computational overhead, we adopt the bounded-delay function [18], [21] to calculate an approximate minimal t on . As shown in Fig.…”

“…The other parts of our algorithms can remain untouched. Suppose the scheduling policy of earliest deadline first (EDF) is adopted, the service bound for the N event streams is [18]:…”