Maximum Likelihood SNR estimation for asynchronously oversampled OFDM signals

In certain OFDM-based communication systems for which data demodulation is not the final goal, it is often of interest to estimate the signal-to-noise power ratio (SNR) from observed samples that are taken asynchronously. Examples include link quality monitoring in broadcast repeaters and spectrum sensing in cognitive radio systems. We examine the structure of the Maximum Likelihood estimate for this problem and propose an iterative method for its computation. The resulting estimate, based on the Karhunen-Loeve transform (KLT) of the data vector, is well behaved and robust to multipath. The computationally intensive KLT can be substituted by the more efficient FFT, asymptotically achieving the same performance.