Potential Research Projects
This page describes potential undergraduate research projects in the SPAN Lab. These are open projects which have potential research impact which have not (yet) been attempted by other SPAN researchers. This is not an exhaustive list. For more detail on any of these projects, please contact Dr. Neal Patwari.
Modeling Inequity as a Feedback System
This project develops tools for analyzing automated decision systems (ADS) which make decisions about what resources people are allocated. Such systems are deployed in a society which already contains inequities and operate in feedback loops in which the outputs of a decision system then impact its future inputs. We note that algorithms which impact the allocation of resources among people are indistinguishable from “policies”, in fact many government and corporate policies are described as algorithms to lend them more credibility. Thus algorithms can be judged to either promote and sustain equity, or to promote inequity, as advocated by Kendi for the antiracist / racist case. In order to analyze policies and ADS, we need to be able to predict how the system will operate in situ in society. Students in this project will model the sum effect of society with a feedback system, and model its predictive ability on temporal data sets.
- Pegah Nokhiz, Aravinda Kanchana Ruwanpathirana, Neal Patwari and Suresh Venkatasubramanian, Precarity: modeling the long-term effects of compounded decisions on individual instability, AAAI/ACM Conf. on Artificial Intelligence, Ethics, and Society (AIES 2021), 20 May 2021. arXiv:2104.12037 [cs.AI]. (Recorded presentation).
Outdoor RSSI data mining
Our lab has developed considerable expertise in environmental monitoring (localization, breathing and pulse rate monitoring) using RF measurements. We are now deploying tens of remotely accessible software-defined radios (SDRs) on towers and building tops in Salt Lake City, Utah. These can be controlled from anywhere and used to emulate current or next-generation wireless protocols. This project is to model the radio channel measurements for the purpose of predicting future values. The model may include current and past RX power, as well as rain, wind, humidity and temperature. Radio channel measurements might include signal strength, Doppler, and/or channel impulse response. These measurements will change over time as the environment changes, for example, when the weather changes. This project is to characterise what kinds of changes are observed as a function of weather, and perhaps other environmental variables.
Relevant Funded Projects:
- Peter Hillyard, Anh Luong, and Neal Patwari, Highly reliable signal strength-based boundary crossing localization in outdoor time-varying environments, ACM/IEEE Information Processing in Sensor Networks 2016, April 2016.
Privacy-preserving transmitter source localization
This project is in support of our NRDZ projecct, which will enable future national radio dynamic zones (NRDZs). These NRDZs could enable more dynamic spectrum assignment in particular geographical areas, and spur faster implementation of new wireless technologies and services. The overall project will develop technologies to enable such zones with monitoring, measurement, and modeling systems which ensure that the new spectrum use in the NRDZ doesn't interfere with other systems outside of the NRDZ. This project will develop privacy-preserving means for the system to accurately learn about a transmitter (its location, transmit power, antenna pattern) so that the system can quantify the interference it offers to systems outside of the NRDZ.