|
Project Summary
Q0S-based
Resource Allocation in Real-time Distributed System
Quanitized
Earliest-deadline-first (EDF) Scheduling
Session-Coordinator
(SesCo)
|
•
Quantized
EDF Scheduling
Model
In EDF, each job has a relative deadline Dj, the current time is t, and its
deadline dj=t+Dj by which the job is inserted into a priority queue.
However, in the environment of limited priority levels, Dj is quantized as
Q(Dj). Its quantized deadline becomes t+Q(Dj) by which the job is
scheduled with EDF.
The above model is analyzed under the framework of Real-Time Queueing Theory
(RTQT).
 |
RTQT
was introduced by Lehoczky and is used to model real-time
tasks under the assumption of heavy traffic stochastic tasks
(with traffic parameter q)
and deadline D with distribution G.
It assumes stochastic arrival (such as mean rate l, with
std of sa
), and service ( such as mean rate m,
with std of ss).
With heavy traffic methodology, we determine the lead-time
profile and lateness in a predictable form.
Major
results include Earliest Deadline First, Quantized Earliest
Deadline First, First In First Out, etc.
|
|
Q-MARS
