Hybrid-Learning-Based Blind Spreading Code Estimation for DSSS Signals in Satellite-IoT Systems

The Internet of Things (IoT) supported by satellites is becoming indispensable for remote sensing in the forthcoming sixth-generation (6G) communication network. However, it is tempting and easy for unauthorized users to exploit the Direct Sequence Spread Spectrum (DSSS) technique to quietly complete their own transmissions due to the open nature of propagation and the publicly available satellite orbits and frequencies. Therefore, it is necessary to conduct a blind estimation of the DSSS signal to take a more proactive approach to protect satellites against illegal use. However, the modulation information is unknown, and the spreading code structure varies, making the blind estimation of spreading codes a significant challenge. Additionally, under data modulation, the dimensionality of the received spread spectrum sequence increases, greatly raising the complexity of spreading code estimation. Against this background, we propose a hybrid learning-based blind estimation algorithm for spreading codes, which combines K-means clustering and Convolutional Neural Networks (CNN). This algorithm achieves low-complexity blind estimation of spreading codes with unknown modulation information and low signal-to-noise ratio. Specifically, the K-means clustering algorithm uncouples the data modulation from the spreading code, reducing the dimensionality of the estimation process. On this basis, the CNN-based parallel convolution architecture is employed to achieve low-complexity and accurate estimation of the spreading code. Simulation results demonstrate that our proposed algorithm outperforms existing algorithms in both computational complexity and estimation performance.