Multi-modal Data Analytics for Tactile Internet
Latency-reliability analysis in statistical sense along with latency reduction techniques to meet the strict QoS requirements of diverse data profiles (audio, visual, haptics) present in multi-modal data, generated in TI applications using MATLAB.- Analyzed latency-reliability for audio, visual, and haptic data streams to meet strict QoS requirements in Tactile Internet applications.
- Developed MATLAB-based simulations to evaluate latency and reliability trade-offs.
- Implemented parallel compression and transmission techniques, achieving a latency reduction of up to 36%.
- Authored a paper accepted at IEEE WCNC 2025 based on these findings.
May 2024 - June 2025 SERB, DST, IndiaCochlear Designing: Auditory system Simulation
Simulated the human auditory pathway in MATLAB using bandpass and gammatone filters to model cochlear response.- Designed and simulated a Digital Human Cochlear System using MATLAB to model the auditory signal processing mechanism.
- Implemented filter banks and signal processing algorithms to mimic the frequency analysis performed by the human cochlea.
- Conducted time-frequency analysis to evaluate the system's performance in accurately decoding auditory signals.
- Utilized MATLAB toolboxes for signal processing and data visualization to validate the model's accuracy and efficiency.
- Documented the findings and presented the results to faculty members, receiving positive feedback for technical accuracy and implementation.
Oct. 2023 - Dec. 2023 IIT Patna, IndiaSign Language Identification
Developed a sign language recognition system achieving 96.5% accuracy on 24 static letters of American Sign Language.- Developed a machine learning model to identify 24 static American Sign Language (ASL) symbols and accurately recognize the corresponding letter.
- Achieved an accuracy of 96.78% using TensorFlow and Keras, with optimized hyperparameters for enhanced performance.
- Utilized Mediapipe for real-time hand tracking and feature extraction, enabling smooth and efficient gesture recognition.
- Implemented data pre-processing techniques such as normalization and augmentation to improve model generalization.
- Conducted extensive model validation and testing to ensure robustness and reliability in real-time scenarios over the constrained devices.
June 2022 - July 2022 BARC Kolkata, IndiaBlood Glucose Level Analysis
Developed a linear neural network to classify sugar into five levels using various biological parameters as input in Python.- Developed a neural network in Python to classify blood glucose levels into five categories based on biological parameters.
- Achieved an accuracy of 95% using TensorFlow and Keras, with optimized hyperparameters for enhanced performance.
- Implemented data pre-processing techniques including normalization and feature selection to improve model efficiency.
- Visualized results using Matplotlib and documented the findings for academic presentation.
Jan. 2023 - May 2023 IEM Kolkata, India