LSTM Networks for Optimal Reference Power AGC in Fading Channel

According to the characteristics of amplitude distribution of communication signals in fading channel, there is an optimal reference power in Automatic Gain Control (AGC) module, which has a maximum signal-to-noise ratio of the output signal for Analog to Digital Converter (ADC). By analyzing the noise power and signal power with ADC sampling, based on the signal amplitude distribution in more general significance, the AGC optimal reference power analysis method establishes the SNR model of ADC sampling output. Taking the outdoor carrier frequency of 17GHz as the analysis object, the optimal reference power analysis method adopts the computer numerical analysis method to obtain the optimal reference power of different ADC parameters in fading channel. The A GC module with the optimal reference power constructs AGC model based on Long Short Term Memory (LSTM) deep learning networks to realize the detection and gain control for fading channel communication signal. The computer simulation result shows that the AGC model has good output stability and the output signal power is stable with the optimal reference power.