Innovations-based channel estimation and symbol detection of multilevel OFDM signals

We propose a state-space model-based approach to the simultaneous estimation of subcarrier fading and symbol quasicoherent detection of multilevel modulated signals in OFDM systems. The receiver uses two low-order Kalman-Bucy filters (KBFs) per mixed subcarrier (data+pilot symbols) and a minimization criterion based on the filters innovations to detect the symbol and estimate the (complex) channel gain. With this scheme we avoid the computational complexity of the conventional minimum mean-squared error solutions. Moreover, since the KBFs estimate the channel continuously no extra time-interpolation algorithm is needed and the amount of pilot symbols can be substantially reduced. Simulation results are presented for 16-APSK in Rayleigh channels affected by residual frequency offset (due to imperfect carrier frequency synchronization).