Signal-to-noise ratio is an important standard to realize adaptive adjustment of parameters in adaptive configuration of LoRa networks. In order to increase the accuracy and stability of SNR estimation for LoRa modulation, an improved SNR estimation algorithm for LoRa modulation is proposed in this paper. On the basis of the SNR estimation algorithm based on spectral analysis and the characteristics of LoRa modulation, the SNR estimation parameter r is defined, and the empirical formula between r and SNR at different SF is determined through experiments, so as to accurately estimate the SNR. The simulation results show that in the best case the SNR that can be accurately estimated by the improved algorithm is improved by -5dB, and when the root-mean-square error of the improved algorithm can reach 2dB the SNR is improved by -14dB. Therefore, compared with the SNR estimation algorithm based on spectral analysis, the improved algorithm has higher accuracy and stability.