Date Added: Dec 2009
The authors consider superimposing pilot symbols on to data symbols for channel estimation for Orthogonal Frequency Division Multiplexing (OFDM) systems. They first derive Maximum-Likelihood (ML) and Minimum-Mean Square Error (MMSE) iterative channel estimators. Modeling the time domain signal as Gaussian, they derive an ML channel estimator by averaging the likelihood function for both data and Channel Impulse Response (CIR) over the resulting Gaussian vector. Two data detectors are also proposed by eliminating the CIR from the likelihood function. The resulting integer least squares problem can be efficiently solved using a Sphere Decoder (SD). Furthermore, the Cramer-Rao Bound (CRB) for the superimposed channel and data estimation is derived.