Differential Ridge Regression-Based Spectrum Map Fusion Under Strongly Correlated Spectral Data

Due to the increasing demand for the accuracy of spectrum maps, fusing spectrum maps has gained attention as an effective method to improve the exactitude of spectrum map construction. However, most of the existing spectrum map fusion methods overlook the over-fitting problem and the correlation of spectral data in the fusion process, so the performance can hardly meet expectations. In this paper, a spectrum map fusion method based on differential ridge regression is proposed, which can construct accurate spectrum maps in the electromagnetic environment with strong-correlation data with high accuracy. First, we construct a spectrum map fusion model by exploiting the propagation characteristics of the spectrum signal. According to the path loss model, the differential ridge regression regularization term is designed to handle the correlation of spectral data and suppress anomalies from spectrum receivers. Finally, we construct a convex optimization problem for spectrum map fusion and obtain the lower bound of the problem by developing Lagrange duality. This method can ensure the convergence of the spectrum map fusion problem and accelerate the convergence speed under low complexity. Simulation results show that the proposed fusion method can effectively improve the accuracy of spectrum map construction compared with the state-of-the-art.