Signal-to-noise ratio estimation method based on deep learning by using constellation diagram

• Main Authors: XIE XIAOJUAN, NI YANQIN, PENG SHENGLIANG
• Format: Patent
• Language: Chinese; English
• Published: 28.04.2020
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention discloses a signal-to-noise ratio estimation method based on deep learning by using a constellation diagram. According to the scheme, the method comprises the following steps: a) generating N digital signals with known signal-to-noise ratios in a [m, n] range, preprocessing the digital signals to obtain corresponding N constellation diagrams, performing data calibration and data setdivision on the N constellation diagrams, and feeding data to a pre-configured deep neural network for training to obtain a deep neural network model M; b) for the received kth observation signal yk,obtaining a kth observation signal yk; preprocessing for s to obtain a constellation diagram with the same size; feeding to a deep neural network model M; model testing, according to the method, the characteristic that the constellation diagram can completely and clearly reflect the signal-to-noise ratio information of the digital modulation signal is fully utilized; a constellation diagram is used as a representation form