Completed work

Completed work and outcomes

Sadly, most of my work (as with many other researchers) has led nowhere... As Lord Kelvin said, “One word characterises the most strenuous of the efforts for the advancement of science that I have made perseveringly during fifty-five years; that word is FAILURE.”

2024

• Developing a synthetic forest dataset for accelerating the training of AI algorithms for forest monitoring (with Frank Feng and Yihang She)

• Monitoring wild boar and lynx population in the Carpathians (with Tom Ratsakatika and Ruben Iosif (carpathian.org))

• Insuring against variability in the performance of Nature-Based climate solutions (with E-Ping Rau, David Coomes, and James Gross, KTH)

• Heatalyzer tool for quantifying the impact of extreme weather impacts on UK housing stock (with Livia Capol and Zoltan Nagy)

• Workload generation tool for EV arrivals and departures (with Anais Berkes)
  • A. Berkes and S. Keshav, SPAGHETTI: A Synthetic Data Generator for post-Covid Electric Vehicle Usage, To Appear, Energy Informatics, 2024.

• Estimating the carbon impact of tropical forest degradation (with Amelia Holcomb and David Coomes)
  • A. Holcomb, P. Burns, S. Keshav, and D. Coomes, “Repeat GEDI footprints measure moderate-scale tropical forest disturbance,” Remote Sensing of the Environment, 308, May 2024.

• Making forest plot measurements faster using a mobile phone and SfM (with Amelia Holcomb, Frank Feng, and Miranda Xie)

• Studying the impact of system configuration on the energy usage of LLMs (with Grant Wilkins and Richard Mortier)
  • G. Wilkins, S. Keshav, and R. Mortier, Offline Energy-Optimal LLM Serving: Workload-Based Energy Models for LLM Inference on Heterogeneous Systems”, Proc. HotCarbon Workshop, July 2024.
  • G. Wilkins, S. Keshav, and R. Mortier, Hybrid Heterogeneous Clusters Can Lower the Energy Consumption of LLM Inference Workloads”, Proc. EEDC workshop at ACM eEnergy 2024, June 2024.

• Joint optimal sizing, and operation of PV, storage, and bi-directional EVs (with Anais Berkes)
  • A. Berkes and S. Keshav, SOPEVS: Sizing and Operation of PV-EV-Integrated Modern Homes, Proc. ACM eEnergy 2024, June 2024.

• Identifying land use from spatial-spectral-temporal remote sensing using self-supervised learning (with Madeline Lisaius, Clement Atzberger, and Andrew Blake)
  • M. C. Lisaius, A. Blake, S. Keshav and C. Atzberger, "Using Barlow Twins to Create Representations from Cloud-corrupted Remote Sensing Time Series," in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, doi:10.1109/JSTARS.2024.3426044

• Using CNNs and computer vision for low-cost and realtime insect biodiversity measurement (with Sachin Matthew, Matteo Redana, and Lynn Dicks)

• Studying how urban buildings can cope with extreme weather events (with Livia Capol, ETH and Zoltan Nagy, UT Austin)

2023

• Tropical Moist Forest Avoided Deforestation Methodology (with Tom Swinfield, Patrick Ferris, Michael Dales, and Robin Message)
  • Balmford, A., Coomes, D., Dales, M., Ferris, P., Hartup, J., Jaffer, S., … Wheeler, C. (2023). PACT Tropical Moist Forest Accreditation Methodology v2.0. Cambridge Open Engage. doi:10.33774/coe-2023-g584d-v5 This content is a preprint and has not been peer-reviewed.

• Exploring the relationship between forest degradation and human development (with Johanna De La Cruz-Rothenfusser and Rachel Garret)

• Inverting radiative transfer models to extract biogeophysical variables from spectra (with Yihang She, Clement Atzberger, and Andrew Blake)

• Using expired weather forecasts to model future weather (with Petr Dolezal and Emily Shuckburgh)
  • P. Dolezal, S.Keshav, and E. Shuckburgh, Using expired weather forecasts to supply up to 10 000 years of weather data, European Geophysical Union General Assembly 2023, Vienna, Austria, 24–28 Apr 2023, EGU23-17356.

• Estimating the rate of carbon sequestration in tropical forests (with Amelia Holcomb, Simon Mathis and David Coomes)
  • A. Holcomb, P. Burns, S. Keshav, and D. Coomes, Computational tools for assessing forest recovery with GEDI shots and forest change maps, Science of Remote Sensing, November 2023.
• Modelling the value of impermanent sequestration (with Tom Swinfield, Andrew Balmford, Ben Groom and Frank Venmans)
  • A. Balmford, S. Keshav, F. Venmans, D. Coomes, B. Groom, A. Madhavapeddy, & T. Swinfield Realising the social value of impermanent carbon credits. Preprint, Cambridge Open Engage, 2023. doi:10.33774/coe-2023-5v93l-v4

• Studying the impact of soiling on solar panels (with Abhinav Bora (Waterloo), Sharat Singh and Hugh Hunt (Quadrical))
  • Data-Driven Estimation of Soiling Loss and Optimal Cleaning Schedule for a Utility-Scale PV Plant MMath Thesis, University of Waterloo, 2023

• Controlling EV charging to allow arbitrary shaping of the load profile of a condominiums complex or a commercial building (with Harsimrat Bhundar and Lukasz Golab, with support from SWTCH Energy, Inc.)
  • H.S.Bhundar, L. Golab, S. Keshav, Using EV charging control to provide building load flexibility. Energy Informatics, 6 , 5, 2023.

• Using ray tracing to optimally site solar panels on rooftops to avoid shadowing and self-shadowing (with Benedictas Tarasovas)

2021

• Estimating the reduction in lighting levels of commercial buildings from Earth observation (with Abhinav Bora)

• Using DNNs to extrapolate Earth observation data to cloudy regions (with Harsimrat Bhundar and Clement Atzberger)

• Privacy-preserving crowdsensing (with Bihan Liu, Steven Chai, and Fiodar Kazhamiaka)

• Using Time-of-flight sensors on mobile phones for depth-assisted segmentation of tree images (with Bill Tong, Megan Penny, and Amelia Holcomb)
  • A. Holcomb, B. Tong, M. Penny, and S. Keshav, Measuring Forest Carbon with Mobile Phones, Poster Proc. ACM Mobisys, Winner of the Best Poster Award, June 2021.
  • A. Holcomb, L. Tong, and S. Keshav, Robust Single-Image Tree Diameter Estimation with Mobile Phones, MDPI Remote Sensing, 15, 772, January 2023.

• Linking Earth observation to carbon sequestration certificates (with Alan Marko, Tom Shan, and David Liu)

• Computing the solar and storage system sizing for the British Antarctic Survey base in Rothera, Antarctica (with Brad Huang, Fiodar Kazhamiaka, and Nopi Exizidou)
  • B.G. Huang, F. Kazhamiaka, and S. Keshav, Sizing Solar Panels and Storage for Multiple Roofs, Proc. ACM eEnergy, June 2021.

• Speeding up Hyperledger Fabric with pre-validation (with Rishav Agarwal, Christian Gorenflo, and Lukasz Golab)

• Implementing RCanopus (with Christian Gorenflo, Ashwin Sekhari, Qingnan Duan, Linguan Wang, Lukasz Golab, Bernard Wong, and (at UMass) Stephen Lee and Prashant Shenoy)

• Creating a trusted marketplace for nature-based solutions (with Tom Shan, Hector Wei, Derek Sorensen, Anil Madhavapeddy)

2020

• Improving the performance of Individual-Based Models for forests (with Siyi Ji)

• Supporting time-based smart contracts in Hyperledger Fabric (with Aritra Mitra, Christian Gorenflo, and Lukasz Golab)
  • A. Mitra, C. Gorenflo, L. Golab, and S. Keshav, TimeFabric: Trusted Time for Hyperledger Fabric, To Appear, Proc. FAB, May 2021.

• Curating datasets collected from ISS4E over the past decade (with Siqing Huo)
  • O. Ardakanian, S. Huo, and S. Keshav, 2020, 6-second load measurement dataset,https://doi.org/10.5683/SP2/R4SVBF, University of Waterloo Dataverse, December 2020.
  • J.Y. Lin, C. Ograda-Bratu, S. Huo, and S. Keshav, 2020, Detecting Snow Covered Solar Panels Dataset, https://doi.org/10.5683/SP2/ITQPHZ, University of Waterloo Dataverse, December 2020.
  • S. Doubov, C. Ograda-Bratu, S. Huo, S. Keshav, 2020, Camera-based Occupancy Detection Dataset, https://doi.org/10.5683/SP2/X30EPQ, University of Waterloo Dataverse, December 2020.
  • Y. Aussat, C. Ograda-Bratu, S. Huo, S. Keshav, "Office Light Level dataset", https://doi.org/10.5683/SP2/3BCADM, University of Waterloo Dataverse, November 2020.
  • A. Rabbani, C. Ograda-Bratu, S. Huo, S. Keshav, "Personal thermal comfort and occupancy dataset", https://doi.org/10.5683/SP2/2NWMDY, University of Waterloo Dataverse, November 2020.
  • C. Gorenflo, I.Rio, L. Golab, C. Ograda-Bratu, S. Huo, S. Keshav, "Webike dataset", https://doi.org/10.5683/SP2/7OAETS, University of Waterloo Dataverse, November 2020.

• Designing a trusted smart meter to ensure that data that enters a blockchain can be relied upon. (with Dimcho Karakashev and Sergey Gorbunov)
  • D. Karakashev, S. Gorbunov, and S. Keshav, "Making Renewable Energy Certificates Efficient, Trustworthy, and Anonymous," Proc. IEEE SmartGridComm, November 2020.

• Estimating IUCN Red List status of birds from Earth observation (with David Wesby, David Coomes, BirdLife International, and the Royal Society for the Protection of Birds)

• Creating a website for homeowners and businesses to determine how many solar panels and how large a storage battery to buy to achieve a certain level of grid independence (with Brad Huang and Fiodar Kazhamiaka)

• Raspberry-Pi system to read analog meters (with Ivy Liu, Yerbol Aussat, and Costin Ograda-Bratu)
  • Github link to source code
• Better teleconferencing (with David Adeboye)
  • Master's report
• Determining when to resize home energy storage systems
  • Bachelor's thesis
  • J. Deller and S. Keshav, "The Benefits of Dynamically Resizing Storage," Proc. Energy Storage Workshop at ACM eEnergy 2020,, June 2020.

2019

• Conflict resolution for hot keys in Fabric
  • C. Gorenflo, L. Golab, S. Keshav, "XOX Fabric: A hybrid approach to blockchain transaction execution," Proc. IEEE ICBC, 2020.

• Using RFID-based sensors for determining soil moisture level
  • J. Wang, L. Chang, S. Aggarwal, O. Abari, and S. Keshav, "Soil Moisture Sensing with Commodity RFID Systems," Proc. ACM Mobisys, 2020.

• Improving the performance of Hyperledger Fabric
  • Gorenflo, S. Lee, L. Golab, S. Keshav. FastFabric: Scaling Hyperledger Fabric to 20,000 Transactions per Second, Proc. IEEE ICBC, May 2019.
• Using Fabric to provide an audit trail for electric vehicle charging
  • C. Gorenflo, L. Golab, and S. Keshav, Mitigating Trust Issues in Electric Vehicle Charging using a Blockchain, Proc. ACM eEnergy 2019, June 2019.
• Using deep neural networks to compute optimal (MPC-computed) storage operation strategies
  • F. Kazhamiaka, S. Keshav, and C. Rosenberg, Adaptive Battery Control with Neural Networks, Proc. AMLIES Workshop at ACM eEnergy 2019, June 2019.
• Using cloud cover and solar panel images to detect panel soiling and shadowing (with Yingjie Chen)

• Using IoT to reduce energy costs of lighting (with Yerbol Aussat)
  • Y. Aussat, A. Rosmanis, S. Keshav, A Power-Efficient Self-Calibrating Smart Lighting System, Energy and Buildings Journal version Author's preprint, January 2022.
• Managing data consent on Fabric
  • R. Agarwal, D. Kumar , L. Golab, S. Keshav, "Consentio: Managing Consent to Data Access using Permissioned Blockchains," Proc. IEEE ICBC, 2020.

• Reducing the cost of personal thermal comfort systems (with Costin Ograda-Bratu)

• Using computer vision algorithms to estimate the degree of snow cover on solar panels (with Jia Ying Lin)

• Improving the accuracy of RFID-based sensors (with Ju Wang and Omid Abari)
  • J. Wang, O. Abari, L. Chang, and S. Keshav, Are RFID Sensing Systems Ready for the Real World?, Proc. Mobisys 2019, June 2019

• Designing and building an off-grid system to capture e-scooter activity and accidents (with Tingyun Liu and Lime Inc.)

• Improving the description of research outcomes from my energy systems research (with Shela Qiu)