Monthly bulletin of the IEEE Computer Society Special Technical Community on Sustainable Computing Providing quick access to timely information on sustainable computing. Sponsored by: ![]() This issue opens with a message from our Chair, which details his plans and goals for the future of the STC-SC community. We then host two technical contributions: Prof. Stelios Krinidis and Prof. Dimitrios Tzovaras (Centre for Research and Technology Hellas/Information Technologies Institute, Thessaloníki, Greece) discuss the adoption of machine learning techniques for building occupancy inference, whereas Dr. Sabidur Rahman (University of California Davis) reports on his recent research about energy consumption in datacenters. As usual, the newsletter closes with the list of upcoming conferences and workshops in the field of sustainable computing. Christopher Stewart Ohio State University, USADear STC Members, Our technical committee has been a driving force for research and engineering in sustainable computing. As the world shifted to cloud computing, our STC fostered research on data center efficiency, renewable powered services and geographic load balancing. As smart grids, smart cities and electric vehicles emerged as key technologies in our sustainable future, our STC sponsored conferences like IGSC and E-Energy, spotlighted researchers and exposed problems in these domain via our newsletter. ![]() The world is changing again. Computer architecture faces the end of Denard Scaling, moving the burden of energy efficiency toward software developers. Smart cities and smart grids are real and increasingly common, but they struggle to efficiently corral and analyze data into actionable insights. Sustainability is bigger than solar or wind energy. Moving forward, sustainable computing solutions must consider long-term effects on society. Are our solutions ethical? Are human, ecological and energy resources sustainable for long-term autonomous deployments? Have we discovered the best, sustainable solutions? We haven't addressed these problems yet. It is fair to wonder aloud. Is our STC ready to address these challenges? It is time to restructure our STC to better support leaders at the forefront of the new challenges in sustainable computing. We will recruit a new board, senior advisors, newsletter contributors and research pioneers. We are looking for fresh faces and established members. This STC must continue to lead by highlighting, promoting and fostering the best research and engineering in sustainable computing. Over the next few months, as we restructure, the newsletter will go dark. In 2019, our STC will reemerge ready to take on new challenges facing sustainable computing. I will reach out to the full membership to plan workshops and request feedback on our restructuring plans. If you would like to be more involved, please email me. We need as many volunteers as possible. I am truly excited about the future for sustainable computing and the important role this STC will continue to play. Thank you, Dr. Christopher Stewart Occupancy Inference based on Machine Learning TechniquesT. Vafeiadis, S. Zikos, D. Ioannidis, S. Krinidis, D. Tzovaras, Centre for Research and Technology Hellas/Information Technologies Institute (CERTH/ITI), Thessaloníki, GreeceThe knowledge of occupancy in domestic environments is vital for many applications, such as energy management, building management, demand/response, security, etc. But, in the majority of the cases, it is very difficult or even impossible to install sensors for receiving occupancy information. Thus, the occupancy (absence – presence or two-class classification scenario) should be detected and inferred utilizing machine learning techniques via electricity and water consumption data received from smart meters in a domestic environment. To this end, the power consumption of crucial electrical appliances (i.e. TV, washing machine, refrigerator, and hair dryer) is monitored every minute. Also, four water consumption sensors have been installed for monitoring the water usage by the occupants in the kitchen as well as the water consumed by the dishwasher and the washing machine. Finally, an occupancy sensor has been installed at the main entrance of the building detecting entries and exits and measuring its occupancy. This information is utilized as ground truth. After retrieving the raw data of the three systems, a processing step was performed in order to create the final dataset which includes events per 1-minute intervals of all the measured features. The initial aggregated dataset constructed after processing the raw data contains 9 features [Central Power (lights of the domestic environment), Refrigerator, TV, Washing Machine, Dryer, Cold Water - Kitchen, Hot Water - Kitchen, Dishwasher - Water, Washing Machine - Water] denoted hereafter as [CP, R, TV, WM, D, CWK, HWK, DW, WMW] and the target Occupancy, denoted hereafter as [OCCUP]. The dataset contains energy and water consumption data of 1-minute resolution for a time interval of 16 consecutive days during summer time. Thus, the shape of overall dataset is 23040x9 (without taking into account the target feature) and its sparsity is 74.44%. In order to rank the influence of each feature to occupancy inference and extract the more useful information, we have used Mutual Information (MI) as the feature selection technique. MI measures how much one random variable provides information. It is a dimensionless quantity, and can be thought of as the reduction in uncertainty about one random variable given knowledge of another. Thus, we decide to use only the top-5 ranked features for occupancy inference, meaning Central Power, Cold Water – Kitchen, Washing Machine, Refrigerator and Washing Machine – Water. Under this condition, the shape of overall dataset is 23040x5 (again without taking into account the target feature) and its sparsity is reduced to 70.76% (from 74.44%). In the sequence, machine learning techniques have been utilized in order to infer the occupancy in a building. The tested machine learning algorithms are:
All these approaches were also combined with the AdaBoost algorithm for even more accurate performance. Our main objective is to find the predictive model that is more efficient on occupancy inference based on energy and water consumption data. To that end, our simulation schema is based on the application of all tested classifiers and their boosting versions on both Initial-DS and MI-DS.
Precision, recall, accuracy and f-measure (estimated averages) for 100 monte-carlo iterations with the application of adaboost, on MI-DS.
Table I presents precision, recall, accuracy and
F-measure (on average of 100 Monte-Carlo iterations) with the application of
boosting on tested classifiers on the dataset. One can see that the DT with
AdaBoost achieves the higher performance compared to the other tested
classifiers (see highlighted values) with 80.94% accuracy (82.31% F-measure),
while the RF follows closely in accuracy (80.23%), but achieves higher
F-measure compared to DT (83.20%). The overall framework has been thoroughly evaluated in [1]. References:
Exploiting temporal and spatial pattern of electricity cost data and dynamic workload migration minimizes data-center electricity costDr. Sabidur Raman, Dr. Abishek Gupta, Prof. Biswanath Mukherjee, University of California Davis, USAProf. Massimo Tornatore, Politecnico di Milano, ItalyFig.
1. 24-hour electricity price for Aug. 25, 2016, from ISOs/RTOs in USA. [3] In particular, as workloads tend to change, often unpredictably, adaptive workload placement and migration can help to serve the workloads in regions with lower electricity costs. Typically, a DC consists of multiple racks, each rack hosts multiple servers, and each server contains many virtual machines (VMs) to serve the client requests. VM migration [6] can be used to exploit the spatio-temporal variation of electricity costs and minimize electricity expenditures. DC workload has dynamic characteristics such as unpredictable workload arrival times and durations, hourly variation in number of incoming workloads, etc. In such cases, quasi-static and batch-processing approaches [4] [5] fail to capture practical aspects of the problem. Instead, a dynamic VM placement and migration approach can capture these practical aspects: electricity cost variation, time-varying workload, and varying duration of services. In addition, our study also considers electricity cost of the backbone network.
Fig.
2. Overview of the problem statement. [3]
Our
findings show promising results on the use of workload-aware VM placement and
migration. A recent article on using DeepMind AI technology to reduce DC energy
cost by 15% shows that a lot more can be done in the field of DC cost
minimization and energy efficiency. Impact of machine learning and data mining
techniques in cloud data center cost minimization and energy efficiency is yet
to be explored in detail. New virtualization platforms such as ‘docker
containers’ are on the rise. Future research can investigate such scenarios. References:
Upcoming EventsThe following venues are requesting submissions on subtopics related to sustainable computing or IT for sustainability.
Conference, Workshop & Symposium Call For Papers
Journal and Special Issue Call For Papers Journal Papers Due Elsevier Sustainable Computing (Open) IEEE Transactions on Sustainable Computing (Open) IEEE Transactions on Green Communications and Networking (Open) IEEE Access, special issue on Social Computing Applications for Smart Cities (http://ieeeaccess.ieee.org/special-sections/social-computing-applications-smart-cities/) Apr. 30, 2018
|
Newsletter >