Privacy Preserving Energy Analytics for Data Centers

Focused Research Program

Our Focus

The goal of the Privacy Preserving Energy Analytics for Data Centers FRP is to demonstrate, for the first time, that large-scale computer systems will achieve cost reduction and energy efficiency improvements via a novel privacy-preserving, collaborative, and scalable analytics and optimization framework. The researchers aim to achieve this goal through a specific high-impact application of our privacy-preserving analytics framework: collaborative data center demand response, where data centers interact and jointly optimize for achieving better electricity cost tradeoffs and energy efficiency in the grid. This research is only possible with a convergent research team and approach, spanning expertise and solutions across energy (power systems and grid), computing design automation, optimization, privacy, computer systems, and hardware. This FRP is cosponsored by Hariri Institute for Computing and the Center for Reliable Information Systems & Cyber Security (RISCS).

Focused Research Program led by:

Research Thrusts

1. Predictive analytics for data center energy trends

This thrust aims to design a collaborative analytics framework for optimizing a federation of data centers to satisfy both individual and joint optimization goals. The framework will predict the power needs that satisfy Service Level Agreements (SLAs), design runtime management policies that are able to meet SLA constraints, and determine which information exchange is necessary among data centers in a federation to jointly participate in demand response programs available in the power grid.

2. Privacy-preserving encrypted analytics

Data privacy considerations, both individual and institutional, hinder collaborations among different data centers (or different users on the grid) or make them vastly more expensive. The aim of this thrust is to design privacy-preserving algorithms that enable sharing power consumption and performance data among data centers for solving joint optimization problems and negotiating joint deals in the power market. The thrust will include:

  • Formalizing the “privacy” requirements, which cover a few distinct categories of concerns (trade secrets, individual privacy);
  • Designing methods that achieve the goals of thrust (a) while satisfying these constraints. These will involve techniques that reduce leakage from the computation’s outputs (aggregation and noise addition, for example) as well as those that reduce the need for pooling data, such as computations on encrypted data.
3. Cross-layer implementation of systems for privacy-preserving analytics

A federation of data centers will jointly solve an optimization problem, which would provide better data center and grid sustainability solutions. To solve these optimization problems, this thrust will explore a variety of approaches including machine learning, simulated annealing, and others, together with privacy-preserving methods such as homomorphic encryption, zero-knowledge proofs, and any new methods designed in thrust 2. This thrust’s goal is to explore hardware design options (i.e., use of accelerators of various kinds, new memory or integration methods, and others) to enable sufficiently fast computation of the privacy-preserving analytics methods.

 

How to get involved?

For program specific inquiries and questions, please contact FRP leaders: Ayse Coskun, Adam Smith, or Ajay Joshi.

Faculty interested in submitting a Focused Research Programs proposal are strongly encouraged to discuss their ideas with Yannis Paschalidis, director of the Hariri Institute for Computing.

To learn more details about the Hariri Institute’s Focused Research Programs, visit here.