My research interests lie in energy-efficient hardware architecture, VLSI design, biomedical signal processing, and machine learning for low-power medical and wearable devices. Our focus is on developing hardware-efficient architectures for digital signal processing, including approximate computing circuits such as multipliers and squaring hardware. In parallel, we work on on-chip machine learning capabilities by designing the VLSI architectures. Ongoing research directions include:
- EEG-based Real-Time Monitoring: Development of systems to monitor the Depth of Anesthesia (DoA) in real-time using spectral and statistical features of EEG.
- Deep Learning for EEG Artifact Removal: Utilizing time-frequency attention transformer encoders and GRU decoders to remove ocular and muscular artifacts from EEG signals.
- Approximate and Neuromorphic Computing: Designing error-aware, energy-efficient logarithmic and antilogarithmic converters for error-resilient digital systems.
Methodologies
VLSI design and synthesis (RTL to GDSII), approximate computing (error-aware design procedures), machine learning for biomedical signals (SVM, K-means, Transformer models), and digital signal processing (denoising and feature extraction).
Applications
Real-time medical monitoring systems, wearable health devices, hardware accelerators for machine learning, and energy-efficient digital signal processing systems.
Domains
Biomedical engineering (EEG, ECG, PPG monitoring), FPGA prototyping, ASIC design and verification.