May 12, 2011 - Available < 80% analyzing.Resource ... '09: Proceedings of the Fourth International ICST Conference on
Supporting Energy-driven Adaptations in Distributed Environments Adel Noureddine, Romain Rouvoy, Lionel Seinturier University Lille 1 & INRIA Lille Nord Europe
12 May 2011
1
Tom watching a movie…
NAS Server
Home Energy Management System
2
Challenges Efficient and consistent adaptations
• •
energy consumption impact on user experience Coordinated adaptations throughout the system
Autonomous adaptations
•
Context evolution energy management no user interventation
Energy modeling
•
Heterogeinity of devices, components, environment
3
Existing works Energy Management
Energy Modeling
Autonomous
SANDMAN [Schiele 2008]
[Binder & Suri 2009]
Transhumance [Paroux 2008]
[Xiao 2009]
PARM [Mohapatra 2003]
Ideal approach
Energy-aware MIDAS
[Petre 2008]
4
Proposed approach Distributed environment Rest-based architecture Autonomic MAPE-K control loop Modeling
• •
QoC dimension: Energy Source Type Resouce abstraction and computation
5
Middleware Architecture
6
Analyzing Component
1. Required resource 2. Available resource
7
Middleware Architecture
8
Planning Component
9
Rules Event Conditions Actions principle Example of a rule used in the prototype for Tom’s scenario Event: Available Resource < Required Resource Condition:
If analyzing.Resource.Available < 80% analyzing.Resource.Required Actions:
execution.Screen.Brightness 70% execution.Sound.Volume 85% execution.Player.Video.Resolution 1024x780
10
Planning Component
1. List of known conflicts 2. Priority system 3. Energy savings estimation
11
Prototype SCA Components component based approach REST principles distributed communications Built on top of the FraSCAti middleware platform [Seinturier et al. 2011]
•
Why? Component based, introspection, heterogeneity, extensibility
Developed (for Tom’s scenario)
• • •
Sensor/Actuator components: Battery, CPU, Screen, Sound and Video Monitoring, Execution and the Decision Engine Knowledge and Rule Respositories
12
Prototype: Video Component Experimentation
d e t la u m
Si
13
Energy (watt)
Simulation (30 min simulated video playback)
Time (minutes)
14
Battery time left (minutes)
Experimentation (39 min video)
Time (minutes)
28
32
15
Discussions Resource
Simulation
Experimentation
Available
Available energy in battery
Battery Timeleft
Required
Energy required for video playback
Video Timeleft
Importance of rules
• •
Stricter rules offer better energy optimization But impact QoS and user experience
Impact of
• •
User preferences Environment
16
Conclusion Insufficient existent middleware solutions Energy-driven middleware architecture Rule-based decision engine REST principles and distributed components
17
Future works (Semi-) Autonomic definition of energy rules
Energy overhead and QoS tradeoff Experimentation in a smart home
18
Thanks
19
References [Petre 2008] Petre, L.: Energy-Aware Middleware. In Proceedings of the 15th Annual International Conference and Workshop on the Engineering of Computer Based Systems (ECBS), IEEE (2008) [Schiele 2008] G. Schiele, M. Handte, and C. Becker. Experiences in designing an energy-aware middleware for pervasive computing. In Proceedings of the 2008 Sixth Annual IEEE International Conference on Pervasive Computing and Communications, pages 504–508, Washington, DC, USA, 2008. IEEE Computer Society. [Binder & Suri 2009] Binder, W., Suri, N.: Green computing: Energy consumption optimized service hosting. In: SOFSEM ’09: Proceesings of the 35th Conference on Current Trends in Theory and Practice of Computer Science, Berlin, Heidelberg, Springer-Verlag (2009) [Paroux 2008] Paroux, G., Demeure, I., Reynaud, L.: A Power-aware Middleware for Mobile Technologies in Distributed Systems (NOTERE), ACM (2008) [Xiao 2009] Xiao, Y., Kalyanaraman, R.S., Ylä-Jääski, A.: Middleware for energy-awareness in mobile devices. In COMSWARE '09: Proceedings of the Fourth International ICST Conference on COMmunication System softWAre and middlewaRE, New York, NY, USA, ACM (2009) [Mohapatra 2003] Mohapatra, S. and Venkatasubramanian, N. 2003. Parm: Power aware reconfigurable middleware. In ICDCS ’03: Proceedings of the 23rd International Conference on Distributed Computing Systems. IEEE Computer Society, Washington, DC, USA, [Seinturier et al. 2011] Lionel Seinturier, Philippe Merle, Romain Rouvoy, Daniel Romero, Valerio Schiavoni, JeanBernard Stefani. A Component-Based Middleware Platform for Reconfigurable Service-Oriented Architectures. In Journal of Software: Practice and Experience, Wiley, 2011.
20