Abstract-This paper presents a second-order blind channel estimation algorithm for digital communication signals under natural asynchronous conditions where no synchronization exists between transmitting and receiving antennas. The approach exploits sampling phase diversity provided by the cyclostationary nature of the received signals as an advantageous means of constructing multiple sets of distinct autocorrelation matrices evaluated at equals time lags. Channel estimation is then performed via joint diagonalization of a set of differential autocorrelation target matrices, avoiding the need of noise power or statistical distribution estimation. A broad set of simulation experiments is presented in distinctive signal contexts as a means of supporting and evidencing the potential of the proposed estimation method.