Improving Symbol Rate Estimation Accuracy by Bandpass Filter Bank

A novel method of using bandpass filter (BPF) bank is introduced in this paper to improve short burst signal's symbol rate estimation performance at low signal-to-noise ratio (SNR) or ratio of symbol energy to noise power spectral density (EsNo), equivalently. In the first stage, we improved the gravity method to have a coarse estimation of the bandwidth of the received signal. And then, a BPF band is designed based on the bandwidth which was estimated in the first stage. The proposed algorithm shows excellent performance compared with other feature extraction methods, which will be verified through simulations. Simulation results also show that the proposed estimation technique can use no more than 200 symbols to achieve a symbol rate estimation accuracy of 3%o at 1dB EsNo for Quadrature Phase Shift Keying (QPSK) modulation.