System Energy Efficiency Lab

System Energy Efficiency Lab
Home People Research Publications Sponsors Contacts

IoT System Characterization and Management: from Data Centers to Smart Devices and Sensors

Power and Performance Modeling and Prediction for Heterogeneous Systems

The emergence of Internet of Things increases the complexity and the heterogeneity of computing platforms. Migrating workload between various platforms is one way to improve both energy efficiency and performance. Effective migration decisions require accurate estimates of its costs and benefits. We are developing a new framework which identifies cross-platform application power and performance at runtime for heterogeneous computing systems. It analyzes and detects machine-independent application phases by characterizing computing platforms offline with a set of benchmarks, and then builds neural network-based models to automatically identify and generalize the complex cross-platform relationships for each benchmark phase. It then leverages these models along with performance counter measurements collected at runtime to estimate performance and power consumption if it were running on a completely different computing platform, including a different CPU architecture, without ever having to run it on there. This work has been supported by Intel Corporation.

Power, Thermal, Reliability and Variability Management for various IoT devices

Power dissipation on integrated circuits makes the temperature increase, which can damage the system and, in the case of mobiles, it can be a source of discomfort for the user. Temperature stress also dramatically increases the impact of reliability degradation mechanisms on transistors and interconnects, which can lead to early failure. These problems only worsen with CMOS scaling, which reduces the accuracy of the fabrication process and increases the variability in power, performance and degradation rate. In our group, we aim at developing and implementing low-overhead and scalable strategies for the joint management of power, temperature, reliability and variability for a variety of devices, from datacenters to mobiles.

Automated Maintenance and Preventive Control of the Internet of Things

The Internet of Things is a growing network of heterogeneous devices, combining commercial, industrial, residential and cloud-fog computing domains. These devices range from low-power sensors with limited capabilities to multi-core platforms on the high-end. The common property for these devices is that they age, degrade and eventually require maintenance in the form of repair, component replacement or complete device replacement. Research in this area approach from a reactive maintenance or preventive maintenance perspectives. In this work, we propose a “maintenance preventive” dynamic control of the IoT devices to minimize the often unforeseen and ignored costs of maintenance. Our work has already demonstrated the importance of dynamic reliability management in mobile systems, by controlling the frequency, voltage and core allocations, while respecting user experience constraints. Our goal is to extend this into the whole IoT domain. We propose optimal control strategies for diverse devices, including sensor devices by adjusting their sampling and communication rates and high-end devices by controlling their frequency and voltage levels. Both solutions are distributed and work towards preventing maintenance costs, while keeping the operational costs to a minimum and data quality within desired limits. We then combine the devices using a smart path selection algorithm that ensures a balanced distribution of reliability across the network. The combined framework works to minimize operational and expected maintenance costs in a distributed and scalable fashion, while respecting the user and data quality constraints imposed by the end-to-end IoT applications.

System Design and Optimization for IoT Ecosystems and Machine Learning

The emergence of IoT increases the complexity and heterogeneity of computing systems. To provide perfect computing environment in this era, we need novel strategies for the system architecture design and system-level optimization. In this context, we have focused on how to analyze, design, and optimize the new systems, with the use of up-and-coming technology such as context/phase-awareness, machine learning, and emerging memory architectures. This helps us better understand system/workload behavior which haven't been seen before, and enables to optimize the diverse devices and systems running over hierarchy. Towards this goal, our recent efforts cover system-level prediction/optimization of cross-platform workload behavior, and new software/hardware co-design for machine learning.

Usage Characterization and Context-Aware Management of Mobile Devices

The popularity of mobiles is increasing up to the point in which there are more smartphones than human beings on the planet. Mobiles today run a variety of interactive applications and embed many different components to provide quality experience to the user. Unfortunately, mobiles deal with problems related to energy efficiency (thus, short battery lifetime), high internal temperature, reliability and variability. In our group we look at the characterization of user interaction to develop reliable benchmarks for mobiles, focusing on the usage of heterogeneous components. We also aim at studying the behavior and requirements of mobile devices in different contexts. Finally, we characterize different phases of mobile applications and implement a framework for their automatic recognition. Our goal is to leverage user and workload characterization to develop and implement a comprehensive context-aware management policy to minimize energy consumption subject to constraints on temperature, reliability, variability and user experience.

Current Members:

Yeseong Kim, CSE PhD, Project: System design and optimization for IoT and heterogeneous systems
Kazim Ergun, ECE PhD, Project: Reliability and maintainability management for IoT hierarchy
Minxuan Zhou, CSE MS, Project: Power and thermal management for smart devices

Alumni Members:

Pietro Mercati, CSE PhD, Project: Variability, Reliability, Thermal and Power Management for Mobile Devices
Wanlin Cui, CSE UG, Project: ML algorithm Characterization for IoT ecosystems
Thomas Worley, CSE UG, Project: HDD power characterization for server systems


  1. Yeseong Kim, Pietro Mercati, Ankit More, Emily Shriver, Tajana S. Rosing, "P4: Phase-Based Power/Performance Prediction of Heterogeneous Systems via Neural Networks", 2017 International Conference on Computer-Aided Design (ICCAD 2017), 2017

  2. Yeseong Kim, Mohsen Imani, Tajana S. Rosing, "CHOIR: General-Purpose Online Classification Accelerator via In-Memory Computing", TECHCON SRC Conference (TECHCON 2017), 2017

  3. Wanlin Cui, Yeseong Kim, and Tajana S. Rosing. "Cross-platform machine learning characterization for task allocation in IoT ecosystems." Computing and Communication Workshop and Conference (CCWC), 2017 IEEE 7th Annual. IEEE, 2017.

  4. Nima Mousavi, Baris Aksanli, Alper Sinan Akyurek, Tajana Simunic Rosing, "Accuracy-Resource Tradeoff for Edge Devices in Internet of Things", IEEE PerIoT, 2017

  5. Yeseong Kim, Pietro Mercati, Tajana Simunic Rosing, "Power Efficient, Hierarchical, Introspection framework for HPC Systems", Techcon, 2016

  6. Pietro Mercati, Tajana Simunic Rosing, "Convex Optimization and Model Identification for Reliable and Energy Efficient QoS-Aware IoT Systems", Techcon, 2016

  7. Pietro Mercati, Vinay Hanumaiah, Jitendra Kulkarni, Simon Bloch, Tajana Simunic Rosing, "BLAST: Battery Lifetime-constrained Adaptation with Selected Target", MOBIQUITOUS, 2015

  8. Pietro Mercati, Francesco Paterna, Andrea Bartolini, Luca Benini, Tajana Simunic Rosing, "Variability Emulation on Real Linux/Android Devices", TECHCON, 2015

  9. Yeseong Kim, Mohsen Imani, Shruti Patil, Tajana S. Rosing, "CAUSE: Critical Application Usage-Aware Memory System using Non-volatile Memory for Mobile Devices", ICCAD, 2015

  10. Yeseong Kim, Francesco Paterna, Sameer Tilak, Tajana S. Rosing, "Smartphone Analysis and Optimization based on User Activity Recognition", ICCAD, 2015

  11. Shruti Patil, Yeseong Kim, Kunal Korgaonkar, Ibrahim Awwal, Tajana S. Rosing, "Characterization of User's Behavior Variations for Design of Replayable Mobile Workloads", MOBICASE, 2015

  12. Pietro Mercati, Vinay Hanumaiah, Jitendra Kulkarni, Simon Bloch, Tajana Simunic Rosing, "User-centric Joint Power and Thermal Management for Smartphones", MOBICASE, 2014

  13. Pietro Mercati, Francesco Paterna, Andrea Bartolini, Luca Benini, Tajana Simunic Rosing, "Dynamic Variability Management in Mobile Multicore Processors under Lifetime Constraints", ICCD, 2014

  14. Pietro Mercati, Andrea Bartolini, Francesco Paterna, Luca Benini, Tajana Simunic Rosing, "An On-line Reliability Emulation Framework", EUC, 2014

  15. Mercati P, Bartolini A, Paterna F., Benini L and Rosing T, "Workload and User Experience-Aware Dynamic Reliability Management in Multicore Processors", Design Automation Conference (DAC), 2013