identifying parameters for code offloading as a practical solution to optimize energy consumption of...
TRANSCRIPT
Identifying parameters for Code Offloading as a
practical solution to optimize the energy
consumption of a smartphone in real life context?
Presentation ON
Outlines
Background
Methodology
Result and Discussion
Conclusion
One of the biggest problems for future growth of
Smartphones is battery life.
Background
One popular technique :
- Remote Execution of code i.e. Mobile Code Offloading
Background
What is Code Offloading:
- Type of binding between mobile and cloud to move a
computational task from one place to another that Increase
performance of the mobile apps.
Background
Code Profiler:
- The code profiler is in charge of determining what to offload.
Background
Code offloading components:
System Profiler:
- Monitoring multiple parameters of the smartphone i.e. energy
to execute the code.
Background
Decision Engine:
- Decide when to offload to the cloud by measuring whether
or not the handset obtains a concrete benefit from
offloading to the cloud.
Background
Surrogate Platform:
- Remote server located in the cloud, which contains the
environment to execute the intermediate code sent by the
mobile device.
Background
Code offloading architecture:
Background
But why it is hard to implement in real life?
Methodology
Parameters for identifying the limitation of
code offloading:
Inaccurate code profiling:
- Difficult to estimate the running cost of a piece of
code considered for offloading.
Methodology
Dynamic configuration:
- This is an evident issue as the computational capabilities
of the latest smartphones are comparable to some servers
running in the cloud.
Methodology
Scalability of the system:
- when the system handles heavy loads of computational
requests its difficult to maintain a certain quality of
responsiveness.
Methodology
Methodology
Characterization of Google App Engine (GAE):
Methodology
Start Quickly, Build Faster:
Methodology
Automatic Scaling:
Methodology
Automated Security Scanning:
Methodology
Tasks Queue:
Characterization of different code offloading
frameworks:
Result and Discussion
Result and Discussion
Comparison among different code offloading frameworks
and GAE (Google App Engine):
0
1
2
3
4
Code ProfilingAccuracy
Acceleration Scalability Dynamic Integration
MAUI
CloneCloud
ThinkAir
COMET
EMCO
GAE
0.0-1.0 Low ; 1.1-2.0 Medium; and 2.1-3.0 High;
Conclusion
My research has explored :
A deep knowledge in the code offloading
architecture and existing different frameworks,
What are the parameters to find out the
limitation while code offloading is working, and
The characterization of Google App Engine as
a solution to overcome those limitation.
So that a developer can come with a suggestion on
which way he/she should choose to develop
mobile Apps to optimize the energy consumption of
a smartphone.