Approximately 10 REU positions will focus on computing and data analysis for an equitable energy transition as described below. 3 REU positions will not be computing focused, but will use a multidisciplinary approach to solving problems in the sphere of equity and the energy transition. These other projects can be seen in the ETI Project Descriptions section below.

Disclaimer: These project descriptions are intended to give you a sense for the work each lab does, exact projects will vary.

Golbon Zakeri- Stochastic processes for wind and hydro adapted to climate change

We have developed stochastic processes models that describe wind speed (and consequently, given a wind plant's specifications, its power output) in the New England region. Similar models have been developed for hydro inflows. In this project, the REU student will examine the goodness of fit of the above developed models to the available data. Furthermore, we will experiment with tuning the model parameters to mimic the effects expected on wind (and hydro) as climate change progresses and atmospheric temperatures rise.

Matt Lackner - Modeling Decarbonization

Project description: Modeling the New England electrical grid. In this project, wind and solar data will be used to develop a tool for modeling the decarbonization of the energy system. The student will analyze resource data, model production, and simulate various scenarios. They will develop a user friendly tool for others to use as well. programming skills, especially Matlab or Python, are needed.

Ramesh Sitaraman - Carbon-aware internet-scale Distributed Systems

The internet is an essential element in the lives of billions of people who use online services to access news, entertainment, social networking, teleconferencing, and e-commerce. These online services are hosted on internet-scale distributed systems that provide content delivery, cloud, and edge computing services that consume increasingly large amounts of energy. This project will look at specific components of these distributed systems and try to understand their carbon and energy footprint with the goal of making them carbon-aware and sustainable.

Jay Taneja - The Equity of Outages

The incidence of electricity outages in the US has been growing for decades. This trend is amplified by infrastructure getting older and climate change increasing the number and scope of extreme weather events. Society already depends on electricity in many ways, and that dependence is only growing as more communities pursue agendas to decarbonize energy systems by converting end-uses to electricity (including transportation, heating, cooking, and more). We ask the question - are these outages felt equally or are they concentrated in low-income or high-minority communities? Using novel datasets of utility-reported outages along with recently-developed methods for estimating the scope of outages, this project will focus on assessing the equity of outage experiences in specific geographies in the US by combining outage data with census data about environmental justice communities.

Erin Baker - Equitable Demand Response.

The objective of this project is to harness the energy transition to improve quality of life for communities that have been historically marginalized and disenfranchised while also addressing the problem of grid integration. Grid integration is the process of incorporating and managing renewable energy sources into the existing electrical grid system. One tool to manage grid integration is Demand Response (DR), which refers to technologies and programs that can change electricity demand patterns to provide a more efficient grid. This REU project will estimate the value of demand response in the economy. The student will use an optimization model to estimate how much energy storage would be needed to match demand to wind and solar energy; the cost of this energy storage provides an upper bound on the value of DR.

Fatima Anwar - Designing more efficient embedded systems (4 possible subprojects below)

Most embedded devices are resource-constrained, and with the proliferation of distributed and learning-enabled edge, we have strict energy and latency budgets. The goal of the following projects is to provide energy-efficient and delay-tolerant distributed learning techniques for emerging applications in 5G and mixed reality tracking.

Domain Adaptive Tracking in Mixed Reality: Mixed reality (MR) applications require tracking systems with high accuracy, low latency, and low jitter. These systems must recover the full six degrees of freedom (6 DOF) of the user’s head pose— the position and orientation of the user’s head in the scene coordinate system. While there are several available solutions for stationary and indoor tracking, a limited volume of solutions exists for mobile and outdoor MR systems. The goal of this work is to “robustify” the sensor fusion algorithms by making the associated deep learning models more explainable. An additional area of exploration is domain randomization to make feature-encoders more robust to environmental noise.

Building an Eye Tracking Module for Mixed Reality Research: Tracking user’s eye fixation direction is crucial to mixed reality (MR). It eases user’s interaction with the virtual scene and enables intelligent rendering for enhancing user’s visual experiences. For conducting research in MR, this project focuses on developing a tool that extracts raw eye tracking data, which is useful for designing robust tracking algorithms.

Defending against timing attacks in 5G communications: 5G distinguishes itself in its support of time-critical emerging IoT applications at a greater scale, hence the focus of this project is on securing Time Sensitive Networking (TSN) over 5G communications. This project will provide secure and reliable TSN for 5G wireless networks in the presence of man-in-the-middle attacks. The design goal is to overcome malicious delays in the network by minimizing delays in one-way packets and making delays symmetric in two-way packet exchange through network protocol and fault-tolerant mechanisms.

Developing efficient distributed learning techniques for multimodal edge: Prevalent distributed learning techniques such as federated learning and split learning preserve data privacy but come at the cost of increased resource consumption and reduced model performance for edge devices with multimodal capabilities. The goal of this project is to develop learning techniques that optimize resource consumption and increase learning performance, and at the same time scalable to thousands of multimodal devices while maintaining privacy.

Jimi Oke - Patterns and predictors of urban emissions

Urban areas account for about 70% of greenhouse gas emissions globally. Cities accounted for 63% of total US emissions in 2012 (Gately et al., 2015). Yet, their patterns and predictors have been relatively understudied. This project will analyze multivariate emissions patterns across US urban areas. We will also develop models to explain these outcomes based on explanatory variables from mobility, socioeconomic, geographic, urban form, environmental, and public health categories. We hope to obtain deeper insights into the drivers of urban emissions and facilitate future equity-aware analyses of climate change mitigation strategies.

David Irwin - Carbon-Aware Job Scheduling

This project would investigate optimal job scheduling to minimize carbon emissions, while maximizing performance. Electricity's carbon emissions vary over time, as the mix of generation sources change. Thus, computer systems can reduce their carbon emissions by running more during low-carbon periods. However, this requires systems to determine when and how many jobs to schedule based on current and future expected carbon emissions. This project will develop policies for determining when and how many jobs to schedule based on predictions of future carbon emissions, which are largely a function of cloudiness (which blocks solar irradiance) and temperature.

Mohammad Hajiesmaili and Prashant Shenoy- Equity Aware Decarbonization and Carbon-Aware algorithms (2 possible subprojects below)

Network- and Equity-aware Decarbonization: The decarbonization of energy systems and associated distribution infrastructure (e.g., natural gas pipelines or electric transmission lines) presents significant technical, social, and economic challenges. As the energy uses currently served by fossil fuels are increasingly electrified, it will be necessary to simultaneously decommission fossil fuel infrastructure and upgrade the electric grid. Additionally, this transition brings natural concerns of equity, as underserved communities have historically received fewer resources and considerations with regard to the placement and condition of energy infrastructure. In this project, we optimize this transition to obtain better socioeconomic and technical outcomes by adopting a network and equity-aware approach.

Applications of Novel Carbon-Aware Algorithms: To design flexible and equitable energy-intensive systems which leverage 24/7 carbon-free electricity, it is necessary to revisit any components of these systems which are incompatible with the characteristics of renewable energy, such as intermittency. This project will investigate opportunities to apply new algorithmic techniques that help systems operate in a carbon-aware and equitable manner.

The deadline to apply is January 26 at 11:59PM ET.

The application will require:

- Personal statement (400-750 words)

- Resume or CV (1 page suggested)

- Transcript (unofficial is fine)

- 1 letter of recommendation

Notifications of acceptance will be sent by the end of February.

In your personal statement we want to get a better sense of who you are, your experiences, and why you are interested in our REU program. We are looking for students who would most benefit from this experience, so be clear how this experience would meaningfully impact your career. Please make this statement no longer than 750 words. Use the following prompts to guide your writing:

• Tell us about your interest/experience in making energy systems more equitable and working with communities on such issues.

• If you are applying to a computing focused project, describe any relevant computing background, such as any programming or data analysis experience, projects, or relevant courses.

• What coursework or other experiences you've had that you think might be related to our research topics?

• Is there anything about your other strengths and interests that you think would contribute to a diverse and fun REU community?

• Connect your interests, experience, and/or career goals to a project you might like to work on this summer.

Examples

The following examples are kindly provided by students who applied to and were accepted into our program.

Resume1 , Resume 2, Resume 3Resume 4

Personal statement1, personal statement2, personal statement3, personal statement4

Climate change is the biggest threat facing both humanity and the natural world. Tackling this threat requires a massive shift in the way we produce and consume energy, and novel computing techniques will play an essential role in optimizing society’s energy- and carbon-efficiency. These new techniques provide a unique opportunity to reshape society towards a more equitable and sustainable state by ensuring that the energy transition benefits rather than harms already marginalized groups. The objective for the Computing for an Equitable Energy Transition (CEET) REU site is to expose undergraduate students to the important and significant role that computing will play in the energy transition, both as an increasingly significant energy consumer and in optimizing society’s energy- and carbon-efficiency in an equitable way. CEET's activities will focus on three distinct computing sub-disciplines that are important for enabling an equitable energy transition, including i) designing energy-efficient, reliable, and low-cost sensors, i.e., sensing, ii) designing energy- and carbon-efficient cloud platforms and applications, i.e., computing, and iii) analyzing collected data to identify and exploit opportunities for improving society’s sustainability that are equitable, i.e., analysis.

The intended impact is to train students to work at the intersection of computing, sustainability, and equity. Sensing-centric projects will focus on inferring useful information from coarse, indirect, and unreliable sensors that enable environmental monitoring without expensive large-scale sensor deployments. Computing-centric projects will focus on optimizing energy- and carbon-efficiency to reduce the environmental impact of large-scale cloud platforms and applications. Data-centric projects will develop methods for analyzing collected sensor data to automatically identify and exploit real-world opportunities for energy and carbon savings. In each case, projects will consider problems through an equity lens to better understand the reasons and consequences of inefficiency. Ultimately, CEET's goal is to provide students a broad background in the many areas of computing research that intersect the energy transition.

Students will be paid a $600/week stipend.

Housing is provided.

~$500-600 of travel expenses will be paid for by the program.

Students will receive a meal stipend of approximately $140 per week.

If you have questions about the REU or application process, please contact Professor David Irwin at irwin@ecs.umass.edu.