Spectral leakage suppression properties of linear and quadratic windowing

It is shown that the leakage suppression properties of segment averaging spectrum estimation methods using linear windows and equivalent quadratic windows are asymptotically the same, under the assumption that segments relatively far apart are uncorrelated. Thus, for large data lengths, one can effectively replace linear windowing by its equivalent quadratic windowing and hope to get similar leakage suppression at a substantially reduced computational cost. A simulation example that supports this conclusion is presented.