Optimal quantization of OFDM at digital IF

This paper considers digital to analog conversion of OFDM signals which have been digitally upconverted to a digital IF (Intermediate Frequency). We examine the effect of clipping and quantization on the received constellations and the uncoded bit error rate (BER) and find that the clipping and quantization can be characterized as a gain and the addition of uncorrelated Gaussian noise. Extensive simulations have been performed to verify this result for BPSK, QPSK, 16-QAM, and 64-QAM with 2 to 14 quantization bits. We also present results showing the optimal quantizer clipping factor and resultant optimal signal to noise ratio for the above modulations and numbers of bits. Finally, we present design curves allowing the selection of the minimum number of bits required to achieve target uncoded BERs showing, for example, that only 6 quantization bits are required to achieve a target BER of < 10^-8 for all of the above modulations.