Task Alloc' In Dist' Embed' Systems

1
CMPE 511COMPUTER ARCHITECTURE

• Task Alloc. In Dist. Embed. Systems
• Murat Semerci
• A.Yasin Çitkaya

2
Task Allocation

• It is process of assigning the task in a way that
  deadlines and precedences of the tasks are
  preserved.
• They are represented as DAGs (Directed Acyclic
  Graph). They show precedence and activities
  between tasks.

3
Embedded Systems

• It is any combination of hardware and software
  systems even involving mechanical parts to
  fulfill a specific function. Unlike
  general-purpose computers embedded sytems are
  designed to performed only interested tasks.
• Most known examples are mobile phones, PDAs, MP3
  players, and sensors (in networks).
• The ones in our interest are the systems
  collaboratively performing tasks in a distributed
  environment.

4
Task Allocation in Embedded Systems

• The major problems in task allocation in
  distributed systems are
• Energy Problem
• Computation/Design Limitation
• Real-Time Processing (Dynamic Tasks Emerging)
• Communication And Communication Channels

5
Genetic Algorithm Based Task Allocation I

• Since most optimal task allocations are hard to
  find, they can not be evaluated in a short time,
  real-time allocation must be evaluated by either
  heuristics or genetic algorithms.
• The only criterion is the task deadline.
• It just deals with objection function information
  (deadline). It works globally from a population,
  not a single point.

6
Genetic Algorithm Based Task Allocation II

• The operations are stochastic, not deterministic
  (Crossover, Mutation)
• The individuals whose utilization is lower than
  or equal to predefined node utilization are the
  fittest (It can not exceed 1)

7
LEneS (Low Energy Scheduling) I

• It is one of first task allocation/scheduling
  algorithms considering energy saving.
• Fundamentally uses DVS (Dynamic Voltage Scaling).
• The tasks can be done on a longer time as long as
  the deadline of the all tasks is not exceeded.
• It is a variant of list-scheduling (assign task
  to the node offering earliest finish time) with
  energy-sensetivity.

8
LEneS II
9
LEneS III

• Take advantage of slack times.
• As long as deadline not exceeded, run on low
  voltages.

10
BEATA (Balanced Energy-Aware Task Allocation) I

• It is a heuristics proposed for soft real-time
  heterogeneous embedded systems.
• It assumes no DVS mechanisms. It assumes distinct
  computation time, communication time,
  communication capacities..vs.
• It assumes all nodes are connected by a
  single-hop wired or wireless network.

11
BEATA II
12
BEATA III

• The total energy consumed by the network is the
  sum of energy consumed by active and idle nodes
  and active and idle links.
• It is a variant of list-scheduling.
• The idea is assigning the task to the node
  consuming less energy among a predefined number
  of nodes completing it fast.

13
BEATA IV
14
Constraints of embedded systems

• Time
• Area (processing power, memory)
• Energy
• Cost
• (market size expected at more than 100 billion
  in 2006)

15
Energy Balanced Task Allocation

• The goal is to balance the energy dissipation of
  the elements during each period of the
  application with respect to the remaining energy
  of elements, such that the system lifetime is
  maximized.
• A single-hop cluster of homogeneous sensor nodes
  connected through multiple wireless channels is
  considered.

16
Clustering
Different ways to cluster a task graph
17
Procedure

• An assignment of all the tasks to onto the
  elements,
• The setting of voltage levels for tasks,
• The scheduling of the computation / communication
  activities, specified by the start and finish
  time of activities.

18
Integer Linear Programming

• computation and communication activities with
  precedence constraints are forced by Constraint
  set 1

19
Integer Linear Programming

• computation and communication activities that do
  not have precedence constraints an extra set
  forced by Constraint set 2

20
Integer Linear Programming

• In Constraint set 3, the system lifetime is
  calculated

Overall ILP formulation
21
Heuristic approach

• Assuming that the voltage levels of all tasks are
  set to the highest options
• In the first phase, the tasks are grouped into
  clusters with the goal to minimize the overall
  execution time of the application

22
Heuristic approach

• In the second phase, clusters are assigned onto
  the elements such that the energy dissipation of
  each element is proportional to the remaining
  energy of the element.

23
Heuristic approach

• In the last phase, the system lifetime is
  maximized by lowering the voltage levels of of
  tasks.

24
Example

• An application example
• Application graph
• b) Time and energy costs for executing tasks at
  voltage levels Vh and Vl

25
Clustering steps for the example
26
Simulation results
Lifetime improvement for small scale problems (3
sensor nodes, 3 voltage levels, 2 channels,
CCR1) a)lifetime improvement achieved by the
ILP-based approach b)perfomance comparison of the
ILP-based approach and the 3-phase
heuristic.
27
Simulation results
Lifetime improvement of the 3-phase heuristic for
large scale problems (10 sensor nodes, 8 voltage
levels, 4 channels, 60-100 tasks) a)lifetime
improvement vs. system utilization (u) and the
communication to computation ratio
(CCR) b)lifetime improvement vs. number of tasks
(CCR4)
28
Simulation results
Impact of variation in number of voltage levels
(10 sensor nodes, 4 channels, 60 tasks, CCR2)
29
Simulation results
Lifetime improvement of the 3-phase heuristic
incorporated with modulation scaling (10 sensor
nodes, 8 voltage levels, 4 channels, 3 modulation
levels, 60 tasks) a)small energy/time ratio for
communication activities (Ci10-7) b) small
energy/time ratio for communication activities
(Ci10-6)
30
Simulation results from real world

• Lifetime improvement for the matrix
  factorization algorithm (10 sensor nodes, 8
  voltage levels, 4 channels, 3 modulation levels)
• u0.5
• u0.8

31
Simulation results from real world

• Lifetime improvement for the FFT algorithm (10
  sensor nodes, 8 voltage levels, 4 channels, 3
  modulation levels)
• u0.5
• u0.8

32
Conclusions

• Task allocations are design to fulfill different
  constraints (mostly time, but also energy).
• Optimal solutions due to including integer
  programming can not be found in a realistic time.
  Because of this, heuristics or genetic algorithms
  are used.
• Energy constraint is the crucial constraint in
  embedded system. Since energy can be saved by
  using low power during computation, there is a
  trade-off between time and energy.
• Heuristics for energy savings are either based on
  individual power consumption or on network
  (total) energy consumption.
• Different heuristics are suggested, which depend
  on different assumptions (homogenous or
  heterogeneous...)
• Any other types energy saving methods can be
  proposed based on new solutions(e.g. new
  communication schemes)

33
References

• B. Korousic-Seljak, J.E. Cooling, Optimization
  of Multiprocessor Real-Time Embedded System
  Structures, Proceedings, 7th Mediterranean
  Electrotechnical Conference, Vol.1, pp.313-316,
  12-14 April 1994 .
• F. Gruian, K. Kuchcinski, LEneS Task
  Scheduling for Low-Energy Systems Using Variable
  Supply Voltage Processors, Proceedings, Asia and
  South Pacific Design Automation Conf.,pp.
  449-455, 30 Jan.-2 Feb. 2001.
• T. Xie, X. Qin, M. Nijim, Solving Energy-Latency
  Dilemma Task Allocation for Parallel
  Applications in Heteregeneous Embedded Systems,
  Proceedings, Inter. Conf. on Parallel Processing,
  pp. 12-22, 14-18Aug. 2006.
• Y. Yu, V. K. Prasanna, Energy-Balanced Task
  Allocation for Collaborating Processing in
  Wireless Sensor Networks, Mobile Networks and
  Applications, Vol. 10, pp. 115-131, 2005.
• M. A. Weiss, Data Structures Algorithms
  Analysis in C, 2nd Ed., Addison Wesley
  Longman, 1999.