Wireless channels are characterized by highly dynamic time-varying nature often manifesting non-stationary behavior. In this paper we consider the state of a
wireless channel in terms of the piecewise covariance stationary signal-to-noise ratio (SNR) process and parameterize it using the probability distribution function of the received SNR and lag-1 autocorrelation coefficient of the corresponding autocorrelation function (ACF). In order to discriminate the state of a wireless channel when its characteristics vary in time we apply methods of statistical process control. Particularly, we use exponential weighted moving average (EWMA) change-point statistical test to detect shifts in the mean value of SNR process. The proposed approach is tested using artificial traces and SNR measurements of IEEE 802.11b wireless channel. Obtained results demonstrate that changes in wireless channel characteristics can be timely detected in real-time. The proposed approach is useful in cross-layer environment where higher layers often need explicit information about the current state of the wireless channel.