Automatic Gain Control (AGC) system performance generally depends on the ability to accurately estimate receive signal power. The initial gain setting may be set high to ensure that low power signals can be detected. As the Variable Gain Amplifier (VGA) resides before the Analog to Digital Converter (ADC), there is a potential for heavy saturation at the output of the ADC, where this estimation is normally performed. This heavy saturation will cause power to be underestimated when using estimators that do not account for clipping, causing slow AGC convergence time. Fast AGC convergence time is desirable in general, but especially for high rate packet based systems (e.g. 802.11ad) which allow for only a limited number of updates in the early stages of a training period. Furthermore, optimal setting of the VGA gain requires statistics sufficient to estimate the received signal distribution. In this work, we propose Maximum Likelihood Estimation (MLE) based estimators of the signal power and noise power that account for saturation at the ADC. The performance of MLE based estimators of total power is compared to a traditional power estimator and results are shown through simulations.
