Unbiased Scattering Function Estimators for Underspread Channels and Extension to Data-Driven Operation

We propose two methods for the estimation of scattering functions of random time-varying channels. In contrast to existing methods, our methods exploit the underspread property of these channels to achieve good estimation performance and low computational complexity. The first method uses a dedicated sounding to measure the channel. The second method uses the data signal of an ongoing data transmission as sounding signal and thus allows estimation without dedicated sounding. Both methods are effectively unbiased and can be implemented efficiently using the Zak transform. The performance of our scattering function estimators is studied both analytically by means of variance bounds and experimentally through numerical simulation, and their superiority over existing methods is demonstrated.