Symbol Rate Estimation Based on Deep Learning.

Satellite signal symbol rate estimation is an important problem in spectrum sensing. Once the symbol rate is known, the received signal can be re-sampled to match the symbol rate. Thus accurate symbol rate estimation is critical for performance optimization of communication systems, and is beneficial for real-time applications (e.g., situational awareness) that require fast analysis of large bandwidths. To this end, this paper proposes a deep learning based symbol rate estimation method, which predicts the symbol rate through utilizing the feature vectors of spectral signals. Experimental results show that the proposed method can achieve a higher estimation accuracy than than the traditional methods.