Effect of average power estimation error on adaptive MQAM modulation

We consider the effects of imperfect average power measurements on adaptive MQAM modulation, where the transmit power and data rate are varied relative to the received signal power. The channel varies with both fast Rayleigh fading and slow log-normal shadowing. We assume that the fast fading is estimated perfectly, and that the estimation error of the shadowing is log-normally distributed. This estimation error leads to a change in the average transmit power and rate of the adaptive modulation. We characterize these changes for two adaptive modulation schemes: variable-rate variable, power MQAM, where the average data rate is maximized, and fixed-rate MQAM, where the transmit power is adapted to invert the signal fading. The shadowing estimation error affects these two modulation techniques in opposite ways: the data rate and dB power changes resulting from the error are positive for variable-rate variable-power MQAM, and negative for fixed-rate MQAM. In both cases, however, the changes are very small.