System Energy Efficiency Lab
Home People Research Publications Sponsors Contacts
Current Research 
Past Research 

Current Research:

Temperature Modeling and Management for Multiprocessor Systems

Thermal hot spots and temperature variations have brought new challenges in reliability, performance and cooling costs in deep submicron system-on-chips (SoCs). Our goal is to develop adaptive low-cost techniques for managing temperature. In addition to providing an optimal solution for minimizing and balancing temperature on the die, we proposed both reactive and proactive dynamic job scheduling techniques. In our work, we have provided extensive experimental evaluation using real life workloads and systems. Dynamic thermal management techniques require monitoring system characteristics (i.e., temperature and workload dynamics) at runtime. Temperature sensors typically are prone to inaccuracy, and some systems lack a sufficient number of sensors on the die. To address these issues, we developed novel solutions for efficient sensor allocation and placement, and for indirect temperature sensing to estimate temperature based on a limited number of noisy sensors.

Data Center Energy Efficiency

The cost of energy consumption in modern data centers has reached and even surpassed the cost of the physical data center itself, necessitating research for dynamically reducing the amount of energy used for computing, cooling and maintaining a data center. The primary goal of this work is the development of a data center power management scheme that delivers energy efficiency with minimal impact on performance. The scheme is based on developing policies for power management techniques like dynamic power management (DPM) and dynamic voltage frequency scaling (DVFS) using online learning and dynamic workload characterization, where the policies adapt to changes in the workloads. Our experiments with CPUs and hard disks confirm the efficiency and adaptability of our online learning algorithms. We further propose extensions to adapt this approach to a virtualized environment encompassing multiple virtual and physical machines, characterizing virtual machines at the hypervisor level to drive both the power management policies and energy aware scheduling. The energy aware scheduler should schedule virtual machines both within and across physical machines for higher energy efficiency.

Wireless Healthcare

A key challenge observed during the clinical trials of preventive mobile healthcare systems was extending the battery lifetime of the system. We analyzed that in such systems with processing, sensing, and communication tasks, it is difficult to dynamically allocate and schedule these tasks on the devices. However, how we allocate and schedule tasks becomes very important if we want to achieve a good tradeoff between battery lifetime and high fidelity of sensed and processed data. Since task binding optimization has been shown to be NP complete, we use an optimal and static ILP formulation of this problem as a baseline comparison to our own dynamic heuristic algorithm which executes at system run time with very low overhead. In a past project, a reconfigurable embedded sensor node, SunSPOT, was shown to significantly extend its set of capabilities through run-time reconfiguration. In this project EKG patterns are tracked in patients and analyzed. The device can then reconfigure itself as necessary to be able to detect different types of heart conditions at very low cost.

Energy Management in Heterogeneous Memory Architectures

Power consumption is a major concern in the design of modern systems. Recent studies have shown that current main memory system based on DRAM has become a significant energy consumer, contributing to as much as 30-40% of total consumption on modern server systems. Our goal is to employ novel memory technologies such as Phase-change RAM (PRAM) and to architect a heterogenous memory system to optimize performance, cost, and power. While PRAM consumes little static power due to its non-volatility, it introduces new challenges associated with limited endurance and high power cost of write accesses. Our experimental results of a heterogeneous system composing both DRAM and PRAM indicate that our solution is able to achieve energy savings up to 37% at negligible overhead over conventional memory architecture.


Past Research:

Event-driven Power Management

Power management (PM) algorithms aim at reducing energy consumption at the system-level by selectively placing components into low-power states. Formerly, two classes of heuristic algorithms have been proposed for power management: timeout and predictive. Later, a category of algorithms based on stochastic control was proposed for power management. These algorithms guarantee optimal results as long as the system that is power managed can be modeled well with exponential distributions. Another advantage is that they can meet performance constraints, something that is not possible with heuristics. We show that there is a large mismatch between measurements and simulation results if the exponential distribution is used to model all user request arrivals. We develop two new approaches that better model system behavior for general user request distributions. These approaches are event driven and give optimal results verified by measurements. The first approach is based on renewal theory. This model assumes that the decision to transition to low power state can be made in only one state. Another method we developed is based on the Time-Indexed Semi-Markov Decision Process model (TISMDP). This model allows for transitions into low power states from any state, but it is also more complex than our other approach. The results obtained by renewal model are guaranteed to match results obtained by TISMDP model, as both approaches give globally optimal solutions. We implemented our power management algorithms on two different classes of devices and the measurement results show power savings ranging from a factor of 1.7 up to 5.0 with insignificant variation in performance.

Energy-efficient software design

Time to market of embedded software has become a crucial issue. As a result, embedded software designers often use libraries that have been preoptimized for a given processor to achieve higher code quality. Unfortunately, current software design methodology often leaves high-level arithmetic optimizations and the use of complex library elements up to the designers' ingenuity. We present a tool flow and a methodology that automates the use of complex processor instructions and pre-optimized software library routines using symbolic algebraic techniques. It leverages our profiler that relates energy consumption to the source code and allows designers to quickly obtain energy consumption breakdown by procedures in their source code.

Energy-efficient wireless communication

Today’s wireless networks are highly heterogeneous with diverse range, requirements and QoS. Since the battery lifetime is limited, power management of the communication interfaces without any significant degradation in performance has become essential. We show a set of different approaches that efficiently reduce power consumption under different environments and applications.When multiple wireless network interfaces (WNICs) are available, we propose a policy to decides what WNIC to employ for a given application and how to optimize the its usage leading to a large improvement in power savings. In the case of client-server multimedia applications running on wireless portable devices, we can exploit the server knowledge of the workload. We present a client- and a server-PM that by exchanging power control information can achieve more than 67 % with no performance loss. Wireless communication represents a critical aspect also in the design of specific applications such as distributed speech recognition in portable devices. We consider quality-of-service tradeoffs and overall system latency and present a wireless LAN scheduling algorithm to minimize the energy consimption of a distributed speech recognition front-end.


An Energy Efficient Routing and Scheduling Mechanism For Ad-Hoc Wireless Networks

In large-scale ad hoc wireless networks data delivery is made challenging by the lack of a network infrastructure and limited energy resources. We propose a novel scheduling and routing strategy for ad hoc wireless networks to address these challenges. Our solution achieves large power savings (up to 60%) while delivering data efficiently. The scheduling algorithm switches off the wireless interface of a large number of nodes for a significant fraction of time thus achieving large energy savings. The algorithm runs above the MAC layer in a completely distributed manner. We test our ideas on a heterogeneous wireless sensor network we have deployed in southern California - HPWREN. Routing relies on a backbone of active nodes that dynamically change over time. The backbone nodes are responsible for delivering the packets to the proper locations. Those nodes that are not part of the backbone run our low-power scheduling algorithm.

SHiMmer: A Wireless, Energy-Harvesting Platform For Structural Health Monitoring

SHiMmer is a wireless platform that combines active sensing and localized processing with energy harvesting to provide long-lived structural health monitoring. SHiMmer uses piezoelectric transducers (PZTs) to evaluate a portion of a structure to determine if damage exists. Unlike other sensor networks that periodically monitor a structure and route information to a base station, our device acquires data and processes it locally before communicating with an external device, such as a remote controlled helicopter. Because SHiMmer receives all its power from solar cells, energy neutrality is essential - the node should not use more energy than it can harvest. We develop algorithms for achieving high performance while maintaining energy neutrality.