Net-Centric Software and Systems I/UCRC A Framework for QoS and Power Management for Mobile Devices in Service Clouds Project Lead: I-Ling Yen, Farokh Bastani, Krishna Kavi Date: April 1, 2010 Copyright © 2010 NSF Net-Centric I/UCRC. All rights reserved. 2009/Current Project Overview A Framework for QoS and Power Management for Mobile Devices in Service Clouds Project Scope: Image capturing Audio capturing … Personal Info DB Facial Image DB Translation Service Project Schedule: 4: Preliminary evaluation Service Cloud 3: Design QPM framework Personal Info DB Translation Service 2: Evaluate service cloud benefit Local Info Service Video DB Facial Image DB Select the services in the cloud To save power on the handheld device Tasks: 1. Develop experimentation environment, including the power measurement, power parameter settings, application suite, etc. 2. Evaluate the potential benefit of delegating service execution to service cloud in order to save power for mobile devices 3. Design QPM (QoS and power management) framework to realize the proposed idea 4. Preliminary evaluation of the QPM framework 1: Enable power measurement and management A M J J A S O N D J F M A 09 10 Deliverables: • Experimental results showing the benefit of using service cloud in saving power for mobile devices • Design of power optimization algorithms Success Criteria: • This project will demonstrate a significant improvement in reducing power consumption on mobile devices 7/12/2016 Page 2 Significant Finding/Accomplishment! Complete Partially Complete 2009 Project Results TASK 1: Develop power measurement and management capability on Linux laptop using ACPI Not Started STAT PROGRESS and ACCOMPLISHMENT Only able to read power consumption every 10 seconds. Only able to get overall power consumption from battery. 2: Evaluate the potential benefit of delegating service execution to service cloud on power saving for mobile devices 3: Design the QPM framework Completed the first version of the design. 4: Preliminary evaluation of the QPM framework Prototype demonstrated with contrived data only. Intermediate results submitted to IEEE SOSE conference. Waiting for IAB approval. 5: Publish results Completed two sample case studies. Results show significant reduction in power consumption on the mobile device. This research illustrates the potential effectiveness of using service cloud for power conservation on mobile devices 7/12/2016 Page 3 Major Accomplishments, Discoveries and Surprises 1. Evaluated the potential benefit of executing services in the cloud for saving power on the mobile device Image Size Platform Selection 100x90 pixels Cloud 50x50 pixels Cloud Latency (Seconds) Comput. 36.80 Comm. 0.07 Local Local 17.80 0.057 Total Local power (watt*sec) 36.87 659.97 (17.9w) 82.70 2315.60 (28.0w) 17.86 289.28 (16.2w) 44.50 1152.55 (25.9w) 2. Designed the QPM (QoS and power management) framework, focusing on two major features • Better prediction. Predicting the workloads and execution patterns of the next time periods • Collaborative. Service cloud performs analysis on historical information and defines parameterized policies. Mobile device makes on-the-fly decisions accordingly 7/12/2016 Page 4 New Problems • Whether to execute a service in the cloud or on the mobile device? • • When it can access service cloud directly When it must access service cloud through other mobile devices • Which services should be allocated on the mobile device? • Minimize the communication cost by bringing services closer to the users Existing works • None consider the same problem • • Existing prediction models are generally simplistic. Hence, QoS and power management decisions can be improved None consider service migration/replication Image capturing Audio capturing … Personal Info DB Facial Image DB Translation Service Service Cloud Personal Info DB Translation Service Local Info Service Video DB Facial Image DB 7/12/2016 Page 5 Our Solution • Better prediction is the key to better power and QoS management • What to predict: • Applications to be executed next • Execution patterns of these applications • How to predict: • For each application, for each specific input, gather their history of events and obtain execution patterns • Consider current and potential future tasks, aggregate their historical execution patterns • Decision process • Offline analysis in the service cloud to determine the best QoS and power management parameters Derive rules accordingly • Mobile device makes on-the-fly decisions based on the rules QPMMD SPESD QPM SADM QoS-PM Service Profile SPESD QoS-PM SADM SMI CAM-AM User Profile Power Manager Service Profile User Profile 7/12/2016 Resource Profile Page 6 2010/New Project Summary A Framework for QoS and Power Management for Mobile Devices in Service Clouds Tasks: 1. Build the experimental environment on Android mobile phone (G1) and PlanetLab 2. Develop the prediction algorithms to predict execution patterns of future time periods 3. Develop the execution decision algorithms 4. Develop the service migration infrastructure 5. Develop the service allocation decision algorithms 6. Validate the framework design Project Schedule: Task 6: Evaluation Tasks 2,3: Simple data collection + coordinated prediction & decision Task 1 Task 6: New Evaluation A M J J A S O N D J F M A 10 Research Goals: 1. Making use of service cloud to improve power saving on mobile devices while satisfying QoS goals Increase the accessibility of the users to complete their critical tasks Increase the situation-awareness and agility of users in special environment (trapped or war fighting) 2. Extend the research to a cloud of mobile devices and Internet-based service cloud Tasks 1-3,6: Enhanced data collection, prediction & decision 11 Benefits to Industry Partners: 1. Significantly improved techniques in QoS and power management 2. Advanced design and prototype framework implementation 3. Experimental results to understand the promising and insignificant factors in QoS and power management 7/12/2016 Page 7