Publikācijas 2007 Multi task real-ime system energy consumption using mini-max method

A. Baums, „Multi task real-time system energy consumption minimization using Mini-Max method” , Electronica ir Electrotechnica, Kaunas, 2007. No. 5 (77), pp.61-65.


Energy consumption reduction is a task of primary
importance in embedded signal processing systems: sensor
networks, mobile robots, wearable computers etc. [1].
Dynamic voltage scaling (DVS) is one of the most
effective energy reduction technology [2,3,4,5,6]. Usually
there are no or few problems if DVS is used in  soft real-times
system design or for task scheduling, where deadlines can be
missed if the Quality of Service is kept [7,8,9].
It is different for  hard real-time  systems where is
necessary to provide that the task never miss its deadline
and the energy saving is maximized. The problems arise in
on-line  systems where the task execution time is unknown
at the starting moment of the ordinary task. Therefore the
best energy reduction regime can not be selected. In
majority of investigations and publications this problem is
solved analyzing periodic task executions with off- line
determined execution times and determined priorities. 
The author has proposed the method of on-line
minimization for the hard task energy consumption in [10].
This method named mini-max is a DVS modification. The
suitability investigation of th is method for different single
task realization regimes was developed in [11]. In this
article the  mini-max method suitability for multitask hard
real-time systems with earliest -deadlines first (EDF) and
rate-monotonic scheduling (RMS) was investigated and the
results are presented.