In this paper, the problem of signal-to-interference plus noise ratio estimation over flat fading channels in non-Gaussian noise is addressed. Most previously published estimators assume the additive noise to be Gaussian. These estimation algorithms performs significantly worse when the additive noise is non-Gaussian. The additive non-Gaussian noise is modeled by a mixture of Gaussians distribution. Both data aided (DA) and non data aided (NDA) estimators are studied. The second and forth order moment based (M2M4) estimator is derived.
Besides, the expectation maximization (EM) algorithm is proposed to iteratively estimate the maximum likelihood (ML) signal to interference plus noise ratio (SINR) estimate for both DA and NDA cases. The performance of the proposed estimators are compared in terms of their mean square error (MSE). Simulation results show remarkable performance improvements of the DA and NDA EM based ML SINR estimator over the M2M4 moment based estimator. Besides, it reveals the degradation in accuracy due to the impulsive noise components over Additive white Gaussian noise (AWGN).