Avhishek Biswas — PhD Researcher in Wireless Systems

About Me

Researcher · Engineer · Builder

I am a PhD researcher in the Cyber Physical Networking (CPN) Lab at the University of Nebraska–Lincoln, advised by Dr. Mehmet Can Vuran. My research focuses on building environment-adaptive wireless systems for 6G networks—from designing RF testbeds and conducting field measurement campaigns to developing channel models and beam management algorithms.

I have collected over 31,600 mmWave propagation measurements across agricultural environments, built wideband channel sounders using USRP X300 radios and Zadoff–Chu correlation pipelines, and developed a camera-primed beam alignment framework (ViBe) that outperforms 5G NR hierarchical beamforming in vehicular experiments. My testbeds integrate 60 GHz phased arrays, sub-6 GHz SDR platforms, and GPU-accelerated signal processing.

I am targeting research engineer and RF engineering roles where I can apply hands-on experience in wireless system design, propagation measurement, and data-driven optimization to real-world communications challenges.

Research Interests

Research Highlights

Research across wireless systems, computer vision, and machine learning

Camera-Primed mmWave Beam Management (ViBe)

Engineered a hybrid closed-loop, hardware-agnostic beam alignment framework for real-time vehicular mmWave communication. ViBe fuses camera-based scene understanding with model-driven beam initialization and online SNR-driven refinement, reducing beam search from O(N²) to constrained local search. Validated across indoor/outdoor testbeds and real-time vehicular experiments, achieving 1.1–1.4% outage rates—outperforming 5G NR hierarchical beamforming and state-of-the-art end-to-end ML models.

mmWave Beamforming V2X Computer Vision 6G

IEEE SECON 2026 (accepted)

mmWave Agricultural Channel Modeling

Led the most extensive mmWave agricultural channel measurement campaign to date, collecting 31,600+ data points in corn and soybean fields across two growing seasons at three farm sites. Developed the mmW-Ag-MLR channel model using multivariate regression with weather parameters (wind, temperature, humidity, vapor concentration, solar radiation) and crop-biological features (plant height, leaf area index). Demonstrated that crop canopy acts as a secondary reflective surface increasing path-loss exponents across growth stages.

Channel Modeling 60 GHz Propagation Agriculture 6G Ag-IoT

IEEE TWC (submitted)

Wideband Rural Channel Sounding

Implemented a correlation-based wideband channel sounding architecture using zero-padded Zadoff–Chu sequences and USRP X300 radios at 5.55 GHz. Conducted LOS and NLOS rural measurements up to 1.3 km at 50 MHz bandwidth (6 m distance resolution, 20 ns delay resolution). Built a complete calibration and signal-processing pipeline: detection, coherent averaging, back-to-back calibration, hardware deconvolution, CIR reconstruction, PDP extraction, and path-loss regression in Python/CuPy.

USRP X300 Zadoff–Chu Channel Sounding Rural Sub-6 GHz

Thermal Hotspot Clustering for Building Energy Audits

Developed a K-means clustering mechanism to identify thermal hotspot regions in building objects (walls, windows) from thermal imagery captured by UAVs. Objects were clustered based on color patterns and pixel temperature distributions. Clustering results were compared against threshold-based approaches, with K-means achieving similar hotspot temperatures at higher spatial density. This work supported scalable, non-destructive energy audit analysis.

K-Means Thermal Imaging UAV Energy Audit

GitHub

DeepThermal: Instance Segmentation for Energy Audits

Contributed to a Mask R-CNN-based instance segmentation pipeline for automated facade and window detection in thermal imagery of buildings. Prepared thermal imagery labels and preprocessed thermal regions to enable automated U-value (thermal transmittance) estimation. The multi-stage approach combined deep learning segmentation with machine learning-based surface temperature prediction for data-driven building efficiency assessment.

Mask R-CNN Instance Segmentation Deep Learning U-Value

Sensors 2021

ML-Based Suicide Risk Prediction

Designed a Bayesian machine learning framework for suicide risk prediction and population-level mental health analytics using Indian demographic data. Built models with Random Forest, Logistic Regression, and Naive Bayes classifiers to identify patterns across social, professional, and educational groups, providing predictions for different risk categories over the past decade.

Machine Learning Classification Scikit-learn Public Health

GitHub

Selected Projects

Systems built for real-world wireless experimentation

Wireless Testbed & Experimental Infrastructure

Designed and deployed mmWave (60 GHz) and sub-6 GHz experimental testbeds for over-the-air wireless measurements. Hardware stack includes phased-array mmWave radios, USRP X300/B200 SDR platforms, external PAs, LNAs, bandpass filters, and GPSDO-synchronized clocking. Software stack includes custom beam-sweeping control, real-time IQ acquisition, GPU-accelerated Zadoff–Chu correlation (Python/CuPy), and automated extraction of SNR, path-loss exponent, RMS delay spread, and beam consistency metrics.

USRP X300/B200 60 GHz Phased Arrays Python/CuPy GNU Radio

GitHub

ViBe: Vision-Based Beamforming Framework

Camera-primed beam alignment framework for vehicular mmWave communication. The system combines computer vision object detection, camera-to-radio geometric coordinate projection, and iterative SNR-based beam refinement to reduce exhaustive O(N²) beam search to constrained local sweeps. Achieves 1.1–1.4% outage rates in real-time vehicular experiments—lower than both 5G NR hierarchical beamforming and end-to-end ML baselines. Hardware-agnostic design requires no offline RF training data.

Python PyTorch MATLAB Wireless InSite Sionna

GitHub

Channel Sounding Signal Processing Pipeline

End-to-end Python framework for processing wideband channel sounding captures from USRP radios. Pipeline stages: signal detection via correlation peak thresholding, coherent averaging across snapshots, back-to-back system calibration, hardware frequency-response deconvolution, calibrated CIR reconstruction, power delay profile (PDP) extraction, and close-in / floating-intercept path-loss regression. All compute-intensive stages are GPU-accelerated with CuPy.

Python CuPy/GPU NumPy Signal Processing

GitHub

Publications

Peer-reviewed journal and conference papers

Accepted

Look Once, Beam Twice: Camera-Primed Real-Time Double-Directional mmWave Beam Management for Vehicular Connectivity

A. Biswas et al.

IEEE International Conference on Sensing, Communication, and Networking (SECON), 2026

Link coming soon

Submitted

Modeling Crop-Specific Environmental Impacts on mmWave Agricultural Channels using Multiple Linear Regression

A. Biswas, S. Nie, M. M. Lunar, G. Bai, Y. Ge, S. Pitla, C. E. Koksal, and M. C. Vuran

IEEE Transactions on Wireless Communications, 2026

Workshop

xGFabric: Coupling Sensor Networks and HPC Facilities with Private 5G Wireless Networks for Real-Time Digital Agriculture

L. Kurafeeva et al.

SC'25 Workshops, 2025

Journal

An Instance Segmentation and Clustering Model for Energy Audit Assessments in Built Environments: A Multi-Stage Approach

Y. Arjouné et al.

Sensors, MDPI, 2021

Paper

Journal

Estimating Surface Temperature from Thermal Imagery of Buildings for Accurate Thermal Transmittance (U-value): A Machine Learning Perspective

A. Biswas et al.

Journal of Building Engineering, 2020

DOI

Journal

Machine Learning Based Prediction of Suicide Probability

A. Biswas et al.

International Journal of Engineering and Advanced Technology (IJEAT), 2020

PDF

View All on Google Scholar

Experience

Research and professional roles

Aug 2021 — Present

Graduate Research Assistant

University of Nebraska–Lincoln · Cyber Physical Networking Lab

Led a two-season mmWave agricultural measurement campaign across three farms, collecting 31,600+ data points in corn and soybean fields at 60 GHz
Engineered the mmW-Ag-MLR channel model using weather and crop-biological variables, improving goodness-of-fit by up to 65% (R²: 0.37 to 0.61) over close-in reference models
Designed ViBe, a camera-primed double-directional beam management framework that reduces beam search from O(N²) to constrained local refinement, achieving 1.1–1.4% outage rates
Implemented a wideband channel sounding architecture with Zadoff–Chu sequences using USRP X300 radios, conducting LOS/NLOS measurements up to 1.3 km
Built mmWave and sub-6 GHz testbeds integrating phased arrays, USRPs, PAs, LNAs, BPFs, and GPSDO synchronization
Developed GPU-accelerated (CuPy) signal processing pipelines for correlation, PDP extraction, and path-loss regression

Sep 2021 — Present

Graduate Teaching Assistant

University of Nebraska–Lincoln · School of Computing

Supported Database Systems, Automata Theory, Computer Networks, and Internet of Things courses for 200+ students
Led labs, recitations, and office hours; designed assignments and rubrics
Mentored students on SQL optimization, network protocol implementation, and IoT project integration

Mar 2020 — May 2021

Junior Research Intern

University of North Dakota · School of Electrical Engineering & Computer Science

Supported development of a machine learning pipeline for UAV-based thermal imagery to automate building heat-loss and U-value estimation
Contributed to Mask R-CNN facade/window segmentation by preparing thermal imagery labels and preprocessing thermal regions with K-means clustering
Enabled scalable, non-destructive energy audit analysis for data-driven building efficiency studies

Technical Skills

Tools and expertise across the wireless systems stack

Wireless Systems

mmWave communication (28–71 GHz)
Sub-6 GHz systems
Beamforming & beam management
Channel sounding & modeling
Propagation measurement campaigns
Vehicular (V2X) communications
5G NR / 6G research

RF Hardware & SDR

USRP X300 / B200 radios
60 GHz phased-array antennas
Power amplifiers / LNAs
Bandpass filters
GPSDO clock synchronization
Over-the-air testbed deployment
Real-time IQ data acquisition

Signal Processing

Wireless channel estimation
Zadoff–Chu correlation
CIR / PDP extraction
Path-loss & delay spread analysis
Coherent averaging & calibration
GPU-accelerated DSP (CuPy)

Programming & ML

Python (NumPy, CuPy, pandas)
MATLAB
C++
PyTorch & TensorFlow
Scikit-learn
Sionna Ray Tracing
Wireless InSite

Tools & Infrastructure

GNU Radio
Linux (Ubuntu)
Git & GitHub
Docker
HPC clusters & Slurm
SSH / remote workflows

Education

Academic background

Ph.D. in Computer Science

University of Nebraska–Lincoln

Sept 2022 — Present

Advisor: Dr. Mehmet Can Vuran
Lab: Cyber Physical Networking (CPN) Lab
Focus: Environment-adaptive mobile wireless systems for 6G, with channel modeling and beam management across rural and urban environments.

B.Tech in Electronics & Communication Engineering

West Bengal University of Technology, Kolkata, India

Aug 2016 — May 2020

Thesis: Bayesian machine learning framework for suicide risk prediction and population-level mental health analytics.
Foundation in signal processing, communication systems, and electronic circuit design.

Service & Outreach

Peer review, mentoring, and community engagement

Peer Review

IEEE Transactions on Wireless Communications (2024–2026)
IEEE SECON (2023, 2026)
IEEE LCN (2025)
IEEE MILCOM (2025)
IEEE DySPAN (2025–2026)

Mentoring

Mentored undergraduate researchers on experimental design, data analysis, and research communication
Guided students on SQL optimization, network protocol implementation, and IoT project integration

Outreach

Presented research projects at Lincoln Hour of Code Fair for undergraduate CS students
Demonstrated research activities at K-12 Outreach Fair

Get in Touch

Open to research collaborations, RF engineering roles, and new opportunities

abiswas3@huskers.unl.edu