Abhinanda Dutta

Experience

• 

  University of Rochester

  Jan 2017 - May 2018

  • Graduate Teaching Assistant

    Jan 2017 - May 2018

  • Graduate Research Assistant

    Jan 2017 - Dec 2017
• 

  Drexel University

  Aug 2018 - now

  Graduate Research Assistant

  1) Applying operations research techniques to solve resource allocation problems in large scale Internet of Things(IoT) networks such as long-range wide area network(LoRaWAN).2) Using combinatorial optimization and stochastic modeling to optimize and establish mathematicalbounds on the performance of LoRaWAN and networks alike.3) Conducting fundamental mathematical analysis on throughput models in random access(RA) scenariounder collision channel model and multi-packet reception model using slotted ALOHA protocol. Show less

• 

  Nokia Bell Labs

  Jun 2020 - Aug 2020

  Research Intern

  1) Implemented an LSTM model to forecast physical resource block(PRB) utilization for network slicing application and the model was tested on a dataset of 1.5 TB, comprising real TCP radio tracesencompassing approximately 7 million rows and 241 features2) Simulated an inter-operator spectrum sharing model using MATLAB and seamlessly integrated it withthe previously implemented LSTM model in Python resulting in an improvement in networkthroughput compared to the previous approach based on naive forecasting. Show less

• 

  Ford Motor Company

  Jun 2021 - Sept 2021

  Research And Development Intern

  1) Implemented and compared ARIMA, SARIMA, LSTM, CNN, and XGBoost models for forecastingQuality of Service(QoS) in Vehicle to Everything(V2X) scenarios using real-time RSSI traces collectedover 8 hours at a millisecond interval across a wide geographic region.2) Developed a Kriging-based model to estimate signal strength at specific locations, facilitating thecreation of a network coverage map.

• 

  Intel Labs

  Jun 2022 - Sept 2022

  AI Research intern

  1) Conducted Spatio-temporal mobile traffic forecasting at the non real time (non-RT) RAN intelligencecontroller (RIC) for network slicing applications.2) Developed a 3D CNN model with custom loss function aligned with network slicing resource allocation policies to predict spatio-temporal mobile traffic within the Telecom Italia dataset, resulting in a36% reduction in service level agreement (SLA) violations compared to the previous naive forecastingfunction.

• 

  聯發科技

  Jun 2023 - Sept 2023

  Research Intern

  1) Designed a Bi-Level Multi-Objective Channel Assignment Problem model (BMOCAP) for integratedsensing, communication and computation (ISCC) system to optimize uplink communication throughput, computational data offloading, and sensing coverage.2) Developed a 2-stage forecasting framework with deep learning and genetic algorithm (NSGA-II) topredict Pareto optimal channel assignments for BMOCAP, resulting in a significant reduction of 48%in generational distance (GD) within the objective space and a 32% reduction in the decision space. Show less