SESSION 7: PRIVACY

This session includes three talks on the privacy issues of energy consumption data collected by smart meters. The first two talks are trying to protect the privacy of energy consumption data with different approaches while the third talk is to show how appliances activities in residential homes can be detected with the modeled energy consumption for different loads.

Talk 1:Shepherd: Sharing Energy for Privacy Preserving in Hybrid AC-DC Microgrids (paper link)

This work investigates the privacy issues in hybrid AC-DC microgrids, where both AC and DC transmission line exist and allow the energy sharing among buildings in the microgrids. Both adversarial models for a single home and microgrids with compromised buildings are proposed for privacy issues. An optimization problem is formulated to minimize the energy transmission in a microgrid while protecting the privacy for all the homes in the microgrid. The results show that the detection ratio with the proposed heuristic solution drops from 33% to 13% compared to battery based consumption hiding approaches.

Q: why there is privacy issue in the proposed setting?

The utility company tries to collect high granularity energy consumption in homes for control or billing purposes. However, they may sell the collected data to third party and the third party may utilize the data to analyze the activity patterns of people for advertisements.

Talk 2:Towards Provable Privacy Guarantees Using Rechargeable Energy-Storage Devices (paper link)

Different from the first talk, this talk is to utilize the rechargeable energy storage devices (e.g., batteries) to hide energy consumption information. The main idea is to change the energy consumption of a load by charging and discharging the battery. The goal of this work is to analyze the privacy guarantees for such strategies. Then a charging strategy is proposed to fulfill the requirement that the charging/discharging rates follow a generalized Irwin-Hall (GIH) distribution. The experiments show that the proposed strategy can give good privacy guarantees and outperforms conventional charging strategies.

Q: The batteries deployed are not supposed for protecting privacy, right? If the batteries is mainly used for protecting privacy, what about the initial purpose?

The work is to investigate the theoretical performance we can achieve if we utilize the battery for protecting the privacy. It can be further discussed how to balance between protecting privacy and other purposes for batteries.

Talk 3:Non-Intrusive Model Derivation: Automated Modeling of Residential Electrical Loads (paper link)

This talk is to provide the insight of complex energy consumption patterns of different appliances in the residential buildings. The loads are categorized into four basic types: resistive, inductive, capacitive and non-linear and the corresponding consumption models for different appliances are proposed. The Non-Intrusive Model Derivation (NIMD) approach is presented for automated modeling of electrical loads: i) when the loads are turned on or off; ii) what are the usage patterns of different loads. The experimental evaluation showed that the automated models are within 1% of the ground truth.

 

Session 6: Green Computing and Networking

Session 6 is about green computing and networking. There are three nice paper presentations in this session.

Paper 1:  Modelling Performance and Power Consumption of Utilisation-based DVFS Using M/M/1 Queues (http://dl.acm.org/authorize.cfm?key=N15719), presented by Prof. Hermann de Meer from University of Passau.

This work uses M/M/1 queues to model the tradeoff between the service performance and the power consumption of utilization-based DVFS. More concretely, under the M/M/1 queuing model for job arrivals and CPU services, the problem is to minimize the mean power consumption of the CPU subject to a mean response time constraint for jobs, and the design spaces are the sample interval size and the utilization threshold to trigger frequency switching. The authors also use extensive simulation to validate their proposed models.

Q1: From the physical point of view, how can voltage be incorporate into the power-frequency model?

Q2: What’s the impact of the size of the sampling interval?

A2: Long sample interval may lead to non-accurate utilization measurement.

Q3: Have you considered the power consumption of non-CPU, e.g. I/O?

A3: Not yet in this work.

 

Paper 2: Energy-efficient Disk Caching for Content Delivery (http://dl.acm.org/authorize.cfm?key=N15725), presented by Mr. Aditya Sundarrajan from University of Massachusetts Amherst.

This work focus on how to reduce the energy consumption of CDNs by shutting down some spinning disks which is used for content caching. The goal of this paper is to reduce the energy consumption of the CDNs but at the same time not to significantly degree the cache performance in terms of hit rate. This paper consider three design spaces: how to determine the cache size, how to shutdown which parts of disks, and how to do content replacement and eviction. It is shown that the proposed solution framework can save 30% disk energy with only a 6.5% decrease in the normalized server hit rate and a 3% reduction in the normalized cluster hit rate.

Q1: What’s the time scale of making decisions in your algorithm?

A1: We adjust the cache size every 6 hours.

Q2 (A follow up question of Q1): This long time scale may use non-updated information and thus degrade the performance.

Q3: How to take into account of smaller content provider and network provider, rather than only considering the energy consumption of CDNs?

A3: It is not considered in the current work.

Q4: Why not use SSD to cache popular contents.

A4: The size of the popular contents is getting bigger and bigger, which may not be cached by SSD.

 

Paper 3: Enabling Reliable Data Center Demand Response via Aggregation (http://dl.acm.org/authorize.cfm?key=N15727), presented by Prof. Yuanxiong Guo from Oklahoma State University.

This work proposes the idea of aggregation to provide reliable demand respond capacity for data centers. It uses a coalition game to model the cooperation of multiple data centers. By elegantly analyzing, the authors first show that indeed when multiple data centers cooperate with each other, the uncertainty of the demand respond capacity can indeed be reduced and the total payoffs of all data centers increases. The authors further propose a payoff allocation scheme to spit the total payoffs into individual data centers, under which all data centers have no incentive to deviate from the grand coalition. Trace driven simulations demonstrate the effectiveness of the proposed approach.

Q1: How to model uncertainly of demand response capacity?

A1: We use a probability density function (PDF) to model it.

Q2: In this model, it seems that it does not have energy reduction?

A2: We do have some discussion in this model, and we may try to do more investigation in the future work.

 

In this session, all three presentations consider very important problems and further show very insightful ideas in the broad category of green computing. I think this is why in this session we have quite active Q&A discussions. Personally, I l favorite the second talk, as I like the nice logics and slides, and I also very enjoy the Q&A discussions, where the presenter and the audiences have a lot of back-and-forth interaction.

Session 5 on Energy-Efficient Transportant

Amazing Session!Very interesting paper and presentations. Blow are some basic information.

#1 Speed planning for solar-powered electric vehicles

Why solar-powered vehicles?

  • 1, Charging time of PEV may be long; Green energy;
  • 2, 22% efficiency of nowadays solar panel, and increasing
  • 3, price of solar energy is reducing
  • Ford C-Max solar energy Concept demoed at CES 2014

However, shading reduces the collected solar power.

Objective: speed planning of solar EVs: ensure enough energy to reach the destination as soon as possible

 

Q: all the information on road is known previously? or you assume there is certain distribution? Yes, here this information is know pre-hand.

Q any particular battery charge/discharge module? Yes, a battery is on the car.

Q: where you get the driving range of these EVs? published.

Q: how big is your battery? 4kWh. Fixed size.

Some random thoughts:

1, The problem formulation: ensure E>0 at the end of each road segment is enough? should it be always >0??

2, What are the shaded and illuminated environments on the experimental paths? how to classify shaded and illuminated in practice?

 

 

#2 Range prediction for electric bicycles

 3-Step approach: get GPS readings; trip segmentation; and battery consumption estimation

The consumption estimation is simply the historic average in the first and simplest model, and, it achieves reasonably well accuracy.

Incorporating more factors in the model actually degrades the accuracy.

Q: why those particular parameters in the 2nd model? Actually, the authors have tried much more.

#3 Fuel minimization of plug-in hybrid electric vehicles by optimizing drive mode selection

Different driving modes (e.g., EV or on fuel) have different energy efficiency. Help drivers to plan the optimal driving modes to save the fuel consumption.

An App on phones, to be connected to the ODB port of the car.

Also incorporated routing planning into the design.

#4 Energy-efficient timely transportation of long-haul heavy-duty trucks

Now comes the best paper candidate!

Heavy consumption (18% of total) of heavy-duty trucks (only 4% in number of vehicles).

Reduce the consumption while ensure latency.

Nice problem formulation presented.

Compared with fastest and shortest approaches, showing significant improvement.

Q: whats the common way to travel? other choices besides shortest/quickest, etc? how they differ from your work?

Q: save 17% consumption? how much additional delay?

 

SESSION 4 Energy Storage

Talk 1:Li-Ion Storage Models for Energy System Optimization: The Accuracy-Tractability Tradeoff (paper link)

This talk is about how to build a model describing the state of charge evolution of Lithium-ion battery, which strike a good balance between accuracy and tractability. They proposed two models and validated by real battery data traces. One of their proposed models works pretty good, in term of tractability (can be easily fitted into a mathematical optimization problem) and accuracy (<5% error). This work is pretty interesting since battery SoC modeling has been a fundamental yet challenging topic in energy storage planning and control area.

Talk 2:Resting Weak Cells to Improve Battery Pack’s Capacity Delivery via Reconfiguration (paper link)

This talk is about how to improve the battery pack’s capacity delivery via  a Cell-Skipping -assisted Reconfiguration (CSR) algorithm. This work is very interesting since the current cell management method, which using all the cells to provide energy, might be in-efficient since the weakest cell dominates their overall capacity. They evaluate the proposed CSR algorithm by large-scale emulation based on empirically collected discharge traces of 40 Lithium-ion cells. And CSR is shown to achieve close-to-optimal capacity delivery when the cell imbalance in the battery pack is low and improve the capacity delivery by up to 94% in case of high imbalance.

Talk 3:Understanding solar PV and battery adoption in Ontario: an agent-based approach (paper link)

 

How to open the door of selling solar PV and battery to customers? This is a very practical yet very important problem. This talk describes how different energy policies affect the adoption of “PV battery systems”. Focusing on Ontario, they conduct a survey to elicit the responsiveness of residents to potential energy policies. And used a Agent-Based Model (ABM) model to forecast the relative performance of different energy policies. The results are very interesting. For one thing, they found PV-battery system adoption in Ontario is likely to be incremental rather than exponential. What’s more, they also mentioned the most effective way to improve PV-battery system adoption is to significantly reduce its price.

Personally, I favorite talk 3. From yesterday’s panel, the question of “how to open the door of smart grid to customers” has been in my mind. Many of us are working on how to optimally do demand response, or how to optimally control EVs. But without customers’ participation, all these proposals can not be finally applied.

The current electricity price is relatively low, therefore “bill saving” might not be good incentive for customers. Prof. Steven Low mentioned yesterday “environmental friendly” might be a good incentive since many people are concerned on our environment. And Talk 3 gave another good point. Reducing the “economy threshold” for customers to participate in the smart grid is also a powerful way to open the door. If the smart meters, solar PV or battery cells are offered at a lower rate (or say free), it might be much easier to help customers get involved into the smart grid.

Session 1

We are pleasured to have Mohammad who is from Chinese University of Hong Kong to present the first full paper in e-Energy 16′.

The title of the paper is “Online Microgrid Energy Generation Scheduling Revisited: The Benefits of Randomization and Interval Prediction”.

As compared to traditional grids, microgrid has recognized advantages in cost efficiency, environmental awareness, and power reliability. In microgrid, intelligent energy generation scheduling is a key mechanism aiming at minimize operation cost and simultaneously satisfy the demand which is composed of electricity and heat.

In this research, the online algorithm is implemented and compared to the performance of the offline algorithm. Two methodologies based on the CHASE are utilized in the research.

  1. Randomized CHASE
  2. Interval prediction type CHASE

Both methods are proven to beyond the best deterministic algorithm. Furthermore, the authors use the electric power data set from San Francisco combining with wind farm generation and a time-of-use scheme electricity price to evaluate their methodologies.

In summary, the paper investigates the potential benefits of randomization and interval prediction in online algorithm design for intelligent energy generation scheduling in microgrid. And their experimental study demonstrates that new design space of randomization and interval prediction can significantly improve the result of the previous deterministic algorithm and can achieve near offline-optimal performance.

 

The second presentation is given by Anamitra who is from IBM Research in India, and the title of the talk is “Supply Scheduling and Usage-Based Pricing for Shared Storage in Adaptive Dynamic Islanding”.

The authors point out that utilities face outages in power distribution due to systemic energy shortfall, device failures, faults in the network, weather condition or intentional electricity outage thus the islanding is a mean to supply backup power to a subset of the load by using local energy sources such as batteries or micro-generation during outages.

The islanding concept is implemented in their study, the main idea in the paper is to share batteries during outages, however, islanding is limited in capacity and not possible to supply all demand. The authors utilized Adaptive Dynamic Islanding (ADI) wherein the utility differentiates between the loads of the customers and supplies the limited energy by cycling it over the customers.

The challenge is that when a battery is used as a shared energy source among a set of customers during an outage, it is necessary to satisfy the demands as much as possible while taking into account the affect of the discharge rate which reduce the battery life.

The authors utilize the data from Irish Commission for Energy Regulation to evaluate their methodologies and the simulations show that their usage-based pricing mechanism (considering battery wear cost) is necessary and is able to capture the varying effects of customers on the lifetime and capacity of the shared battery.

The third talk is also presented by Anamitra, and the title of the paper is “A Network Calculus Foundation for Smart-Grids where Demand and Supply Vary in Space and Time”.

In this paper, the authors point out that energy distribution problem will vary in time and space when electric vehicles (EVs) are involved. The queueing policies which  can be employed to systematically match demand and supply for EVs that capable of swapping the battery.

Two types of policies are used as queueing algorithm for EV to swap battery:

  1. Pure policies which displace demand in only one dimension (time or space):
    1. Higher priority is given to EVs that arrive earlier.
    2. Higher priority is given to EVs that arrive at earlier regions.
  2. Mixed policies
    1. Consider the total demand arriving in a space-window as a single arrival process, and the total supply available in a space- window as a single service process.
    2. Consider the total demand arriving and supply available in a time-window as single arrival and supply processes, respectively.

Quality- of-Service (QoS) metrics are used to evaluate the work. For any such policy, the authors compute QoS metrics, such as upper-bounds on buffer-length, delay, and the output arrival function, for all observable arrival and service functions. The authors design 4 scenario to compare different policies and they demonstrate the methods in analyzing QoS metrics in servicing EVs plying on a highway with batteries charged with solar chargers along the highway. In particular, they consider the metrics of worst-case delay and distance travelled by an EV before be- ing serviced with a charged battery.

Summary of the session:

This session includes three talks focus on the energy distribution in microgrid grid, battery sharing and EV battery swapping. The session covers a wide discussion among the energy distributions.

In this session, I am personally interested in the first talk which considers the online algorithm implemented in the microgrid system. Since it is really difficult to predict the future precisely, therefore, an online algorithm that provides acceptable solution is very practical.

ACM e-Energy 2016 Session 2 (June 22)

Talk 1: Leveraging Energy Storage to Optimize Data Center Electricity Cost in Emerging Power Markets (http://dl.acm.org/citation.cfm?id=2934346)

Yuanyuan Shi made the presentation for this paper. She introduced how to utilize battery storage for participating in the regulation services to reduce the electricity bill. The regulation service is an interaction between the data center and the grid. The cost saving consists of reducing the peak demand charge and gaining revenue from participating in regulation markets. In the trace-driven experiments, the electricity bill can be significantly reduced by the battery-based approach.

 

Talk 2: How to Cool Internet-Scale Distributed Networks on the Cheap (http://dl.acm.org/citation.cfm?id=2934337)

Stephen Lee made the presentation for this paper. He introduced how to utilize the open air cooling and the load-balancing managing in Internet-scale Distributed Network (IDN) for decreasing the total cooling cost in the data center. The cooling cost can be reduced by transporting some workloads among multiple geo-distributed data centers, and thus the joint new techniques can reduce the total cost (i.e., transfer cost and cooling cost).

 

Talk 3: Joint Capacity Planning and Operational Management for Sustainable Data Centers and Demand Response (http://dl.acm.org/citation.cfm?id=2934344)

Tan N. Le made the presentation for this paper. He introduced how to utilize the information predictions to find out a more efficient combination of the capacity planning expenses and operational management expenses in future multiple years. The capacity planning is about how the infrastructures develop in the future years, and the operational management is about the expense of workloads in the future years. Even though the prediction of one source might not be accurate, the framework can still obtain some comparable good results due to the variety of the energy sources.

 

This session mainly introduces three interesting topics in energy-efficient datacenters. The topics include electricity bill reduction by utilizing backup battery, cooling cost reduction by utilizing the open air cooling system, and data center building cost by utilizing the prediction of historical data.

 

Personally, I favorite talk 1 and especially love the trace-driven experiments of electricity markets. It would be better if the future work could prove some theoretical performance bounds or handle the more general battery models.

Welcome to e-Energy 16′ & Keynote

General Chair of seventh ACM e-Energy Dr. Keshav would like to thanks everyone for coming, and hope every enjoy the conference with coming good weather of conference duration. This year there are around 85 attendees in the conference, 24 full papers, 2 keynotes, 4 workshops, 13 posters and 1 panel.

Dr. Steven Low and Dr. Minghua Chen who are Technical Program Committee chairs provide detailed information of the submissions and reviews in the conference. Basically, each paper received 3-5 reviews and the average score is 3.21 out of 6, where the average review score of accepted paper is 3.98.

Keynote and Panel chair Dr. Catherine announces the best paper award this year which goes to the paper titled An online incentive mechanism for emergency demand response in geo-distributed colocation data centers. There are three papers went into the final round:

  1. Online microgrid energy generation scheduling revisited: the benefits of randomization and interval prediction
  2. An online incentive mechanism for emergency demand response in geo-distributed colocation data centers
  3. Energy-efficient timely transportation of long-haul heavy-duty trucks

 

Keynote:

The first keynote is presented by Doug Thomas who is Vice-President of Information and Technology Services, and Chief Information Officer, IESO. The title is Ontario’s Changing Electricity System and the Role of Data. The IESO is the reliable coordinator for Ontario and works closely with other jurisdictions to ensure energy adequacy across North America, and now the number of micro grid contracts has reached 20 thousands.

The change of Ontario is obvious these decades. Ontario has seen the phase out of coal-fired generation, tremendous growth in renewable generation sources at both the transmission and distribution level, and a strong focus on conservation and demand management initiatives. The first wind farm commissioned in 2006 and final coal plan was closed in 2014.

Distributed generation now has material impacts on how Ontario’s power system is managed – $2 trillion is expected to be invested globally over the next decade to modernize grid infrastructure, and the future grid will manage centralize generation in concert with hundreds of thousands of distributed energy resources.

Besides, 21st century customers also change, they expect more from utilities, including clean, reliable, affordable electricity and services, they are shifted form passive to proactive. Therefore, the IESO launch two Demand Response (DR) projects which enable customers to react to the dynamic electricity prices.

To implement DR project, the smart metering system is installed, basically, the system collects the electricity usage data and provides customers the billing information as the feedbacks.It is believed such an intelligent system would help to enhance the energy efficiency of the electric grid and make the environment better. However, the privacy issue is expected and how to working through it will be the next challenge.