Debajyoti Halder

DEBAJYOTI HALDER

Stony Brook, New York    ·    dhalder@cs.stonybrook.edu

I'm a Ph.D. Student in the Dept. of Computer Science at Stony Brook University. I'm working as a Graduate Research Assistant at the PACE Lab. I'm advised by Prof. Anshul Gandhi and Prof. Erez Zadok. My research interests span across Distributed Systems and Peer-to-peer Networks.


I received B.Tech in Computer Science Engineering from the Indian Institute of Technology (IIT) Bhilai in 2022. I have previously worked as a part-time Software Engineer Intern for 2 years at DMLab, Chile.


Experience

  • Reliability of Server Components

    Supervisors: Dr. Anshul Gandhi and Dr. Erez Zadok

    • Reliability: Estimating reliability impact on server components due to workloads.
    • SSDs & HDDs: Built regression models for predicting the Remaining Useful Lifetime (RUL) of SSD/HDDs from SMART values. Models can predict RUL with an MAE of ∼13 days.
    • Per-job RUL impact: Building a model to estimate SMART values based on job profile and component load, and further predict per-job RUL impact on the component.
    • DRAM: Building a regression model to predict RUL of DRAM from DRAM error logs. Training on publicly available error logs datasets.
    • Novelty: Prior works employ classification models to predict whether a component will fail within a certain window, we use regression models to predict per-job impact on the RUL of the component.
  • Per-job Operational and Embodied Carbon Emissions

    Supervisors: Dr. Anshul Gandhi and Dr. Erez Zadok

    • Per-job Power Prediction: Tracking per-job resource usage in servers using perf-stat and blktrace, and predicting power consumption of each workload using ML models.
    • Evaluating ML models: Extensive evaluation of a wide variety of ML models with various power model formulations. The final XGBoost model could predict per-job operational power with ∼7% MAPE.
    • Embodied carbon: Built a model to predict embodied carbon emission from the server configurations.
    • Monitoring tool: Developed a resource monitoring tool to analyze the usage of server components in real-time and report per-job embodied share per-unit time.
    • ACCEPTED at HotCloudPerf 2024: Empirical Evaluation of ML Models for Per-Job Power Prediction - DOI.
  • Research Fellowship from Arista Networks

    Supervisor: Dr. Anand M. Baswade

    • Real-time IDS for WPA3: A two stage architecture for real-time intrusion detection for an enterprise scenario.
    • Attacks on WPA3 testbed: A new, open source dataset for attacks on WPA3-enabled device.
    • ML based IDS: A lightweight ML-based classifier for attack detection with high accuracy in real-time.
    • ACCEPTED at GlobeCom ’22: RIDS : Real-time Intrusion Detection System for WPA3 enabled Enterprise Networks - DOI.
  • Research on SDN in Satellite Networks

    Supervisor: Dr. Anand M. Baswade

    • QoS aware routing: Developed an algorithm better than Dijkstra to find the shortest as well as QoS aware route in a Satellite Network.
    • Protocols for flow rule transfer: Developed protocols for flow rule transfer in SDN which ensure non-disruptive service during satellite handovers.
    • Visualizations: fybrrLink Visualization Tool : fybrrLink-VizTool.
    • ACCEPTED in IEEE TNSM: fybrrLink: Efficient QoS-aware Routing in SDN enabled Future Satellite Networks - DOI.
  • Research on P2P Messaging

    Project Lead

    • fybrrChat: A Distributed Chat Application for Secure P2P Messaging.
    • Distributed technologies: Secure and Instant Messaging using WebRTC and IPFS.
    • Comparison with WhatsApp: fybrrChat is able to perform better than WhatsApp at even lesser cost of operation.
    • Website: fybrrChat test version : fybrrChat-lite.
    • IN REVIEW at GlobeCom ’22: fybrrChat: A Distributed Chat Application for Secure P2P Messaging - Preprint.
  • Research on P2P Live Streaming

    Supervisor: Dr. Anand M. Baswade

    • fybrrStream: A WebRTC based Efficient and Scalable P2P Live Streaming Platform.
    • Low latency: WebRTC based low latency (0.01s) peer-to-peer live streaming.
    • Field test: India-wide field test with 50+ participants spread across 3000+ km.
    • Website: Experiment version : fybrrStream.
    • ACCEPTED in ICCCN '21: fybrrStream: A WebRTC based Efficient and Scalable P2P Live Streaming Platform - DOI.
  • Intern at DMLab, Chile

    Software Engineer Intern

    • Aeris: Live peer-to-peer video calling based physiotherapy
    • Computer Vision: Pose estimation for diagnosis using Mediapipe (Tensorflow).
    • P2P platform: Implemented fybrrChat - WebRTC based P2P video calling for the appointment system.
    • Website: Visit DMLab, Chile.
  • Intern at Elogix Softwares Pvt. Ltd., Kolkata

    Full Stack Development Intern

    • PLMS: RFID based Plant Logistics Management System.
    • Tracking and analysis: Real-time truck inventory issuing, anomaly/theft detection.
    • Data Analysis: Generated custom Jasper reports for Data Analysis.
    • Website: Visit Elogix Softwares Pvt. Ltd., Kolkata.

Publications

  • Workshop Paper - Empirical Evaluation of ML Models for Per-Job Power Prediction

    HotCloudPerf 2024 - 7th Workshop on Hot Topics in Cloud Computing Performance

    D. Halder, M. Acharya, A. Malsane, A. Gandhi, and E. Zadok

  • Conference Paper - RIDS : Real-time Intrusion Detection System for WPA3 enabled Enterprise Networks

    GLOBECOM 2022 - IEEE Global Communications Conference

    R. Saini, D. Halder, and A. M. Baswade

  • Journal Paper - fybrrLink: Efficient QoS-aware Routing in SDN enabled Future Satellite Networks

    IEEE Transactions on Network and Service Management - Special Issue on ‘Smart Management of Softwarized Networks’

    P. Kumar, S. Bhushan, D. Halder, and A. M. Baswade

  • Conference Paper - fybrrStream: A WebRTC based Efficient and Scalable P2P Live Streaming Platform

    2021 International Conference on Computer Communications and Networks (ICCCN)

    D. Halder, P. Kumar, S. Bhushan, and A. M. Baswade

Honors and Awards

  • Alumni Association's Young Researcher Award

    Indian Institute of Technology (IIT) Bhilai

  • Research Fellowship Award

    Arista Networks x IIT Bhilai

  • Honorary Mention Award

    2021 IEEE ComSoc Student Competition

Btw, you can also call me Roton !