Adaptive Signal Detection and Modulation Classification Using CNN and Attention Mechanisms in 6G UAV Systems

Abstract

In this paper, a signal detection and modulation classification network will be suggested based on an Adaptive Enhanced Signal Detection Network (AERSD) in the 6G UAV communication systems. The central problem that it is based on is that the current methods are very complex and inefficient, especially in low signal-to-noise Ratio (SNR) conditions, in which the existing systems are problematic with efficiency and real-time implementation. The AERSD addresses these barriers by applied use of deep learning algorithms, such as an adaptive 1D Convolutional neural networks (CNNs) being used to extract features, SNR-based parameter optimization to extract specific features, and feature weighting through attention allowing it to adapt to challenging and dynamic communication environments. The system is also lightweight and computationally efficient, and thus can be used in real-time UAV communication. The experimental findings indicate that the AERSD model is better than the conventional signal detection techniques and has an accuracy of 93.75, a recall of 0.92, and a detection rate of 0.972. The model also has a low bit error rate (BER) of 0.00118, a low latency of 0.563 ms, and an energy-efficient model. The ratio of throughput and packet delivery was also analyzed under different conditions of traffic, and it was seen that AERSD is strong enough to handle real-time data. Results confirm the AERSD to be an effective, versatile, and computationally economical tool in signal detection and modulation classification in 6G UAV communication systems.