Data-efficient blind OFDM channel estimation using receiver diversity

We investigate non data-aided channel estimation for cyclically prefixed orthogonal frequency division multiplexing (OFDM) systems. By exploiting channel diversity using only two receive antennas, a blind deterministic algorithm is proposed. Identifiability conditions are derived that guarantee the perfect channel retrieval in the absence of noise. In the presence of noise, the proposed method has the desired property of being data efficient-only a single OFDM block is needed to achieve good estimation performance for a wide range of SNR values. The algorithm is also robust to input symbols as it does not have any restriction on the input symbols with regard to their constellation or their statistical properties. In addition, this diversity-based algorithm is computationally efficient, and its performance compares favorably to most existing blind algorithms.