Data Rate Loss Due to Quantized SNR Estimates in Adaptive Bit Loading Algorithms

Abstract

Adaptive bit loading enables OFDM systems to optimize the transmit power levels and/or transmission data rates. These algorithms require the knowledge of ideal signal to noise ratio (SNR) for each data subcarrier at both the transmitter and the receiver. However, in practice the channel SNR can only be obtained and shared between the destination and the information source in a quantized fashion. In this paper, we analyze the effect of SNR quantization on the system data rate. We analytically obtain the average data rate loss due to quantization of SNR estimates. We show, via measurement results on powerline communication channels, that the effect of SNR quantization on the system data rate is not negligible if a midtread quantizer with less than 8 bits is used in the SNR estimation process for powerline communications. We also compare obtained results with theoretical capacity values in order to emphasize the data rate loss due to making use of quantized SNR estimates.