Weak GPS signal tracking using block coherent integration based on NN discriminator

High sensitivity carrier phase tracking is critical for intelligent unmanned mobile devices such as smart drones and autonomous vehicles. To address the issue of sensitivity degradation caused by signal amplitude attenuation in traditional coherent integration (CI) methods, this paper proposes a Blocked Coherent Integration Neural Phase-Locked Loop (BCI-NPLL) method. By constructing a Blocked Coherent Integration (BCI) matrix and integrating it with a hybrid neural network discriminator, the method achieves improved carrier phase estimation. Additionally, a novel sensitivity evaluation approach is proposed to quantitatively assess the sensitivity and accuracy of closed-loop tracking. Through semi-analytical simulations, the loop phase jitter model under the proposed BCI method is validated, and the practical performance of the Neural Phase-Locked Loop (BCI-NPLL) method is tested using both simulations and real GNSS signals. The results show that the Blocked Coherent Integration BCI-NPLL improves the sensitivity by 2 to 3 dB under static conditions compared to traditional PLL methods. Under and acceleration conditions, sensitivity is improved by 4 dB and 5 dB respectively. In addition, BCI-NPLL achieves a to improvement in tracking accuracy as the Carrier-to-Noise Ratio (CNR) decreases.