A Deep Learning-Assisted FPGA Architecture for Real-Time Signal Classification in Next-Generation Wireless Communication Systems

Abstract

The rapid evolution of fifth-generation (5G), Beyond 5G (B5G), and sixth-generation (6G) wireless communication systems has significantly increased the complexity of real-time signal processing due to massive spectrum utilization, heterogeneous modulation techniques, ultra-low latency requirements, and intelligent network management. Conventional software-based signal classification approaches often fail to satisfy the stringent latency, throughput, and energy efficiency requirements of next-generation wireless infrastructures. Deep Learning (DL) algorithms have demonstrated remarkable capabilities in automatically extracting complex temporal and spectral features from communication signals; however, their computational complexity limits deployment in real-time embedded systems. Field Programmable Gate Arrays (FPGAs) provide highly parallel, low-latency, and energy-efficient hardware acceleration suitable for implementing deep neural networks directly at the network edge. This research proposes a Deep Learning-Assisted FPGA Architecture for real-time wireless signal classification by integrating optimized Convolutional Neural Networks (CNNs), hardware-friendly quantization techniques, pipelined processing architecture, parallel inference engines, and FPGA-based acceleration. The proposed framework enables rapid modulation recognition, interference identification, spectrum sensing, and signal classification while maintaining low latency and reduced power consumption. Experimental evaluation demonstrates significant improvements in classification accuracy, inference speed, hardware utilization efficiency, throughput, and energy efficiency compared with conventional CPU- and GPU-based implementations. The proposed architecture provides an effective solution for intelligent edge computing, software-defined radio, cognitive radio, and future 6G communication systems requiring real-time artificial intelligence capabilities