Unbiased estimation of symmetric noncausal ARMA parameters using lattice filter

An unbiased estimation method for symmetric noncausal ARMA model parameters is presented. The proposed algorithm works in two steps: first, a spectral ly equivalent causal system is identified by lattice whitening filter and t hen the equivalent noncausal system is reconstructed. For AR system with no ise or ARMA system without noise, the proposed method does not need any ite ration method nor any optimization procedure. An estimation method of noise variance when the observation is made in noisy situation is discussed. The potential capabilities of the algorithm are demonstrated by using some num erical examples.