Artificial Intelligence Techniques for Deep Learning-based Area Efficient 1024-Point Pipelined Radix-4 FFT Processor for Biomedical Application: Trends and Challenges

Abstract

The increasing demand for real-time biomedical signal processing in applications such as electrocardiogram (ECG), electroencephalogram (EEG), and medical imaging has driven the need for high-performance and area-efficient signal processing architectures. The Fast Fourier Transform (FFT) plays a fundamental role in converting time-domain signals into frequency-domain representations, significantly reducing computational complexity from to . Among various FFT architectures, radix-4 pipelined designs are widely preferred due to their reduced arithmetic complexity and improved throughput. Recent advancements in artificial intelligence (AI), particularly deep learning, have introduced new opportunities for optimizing FFT processor design. AI techniques enable adaptive signal processing, intelligent noise reduction, and efficient resource allocation in hardware implementations. Additionally, pipelined architectures enhance throughput by enabling parallel execution of operations, thereby improving system performance. This paper presents a comprehensive review of deep learning-based, area-efficient 1024-point pipelined radix-4 FFT processors for biomedical applications. It highlights emerging AI-driven optimization techniques, evaluates existing architectures, and identifies key challenges such as power consumption, hardware complexity, and scalability. The study further explores future trends toward intelligent, energy-efficient, and real-time biomedical signal processing systems.