Abstract: The off-the-shelf WiFi network interface card(NIC) can provide channel state information(CSI) which has more detailed information than received signal strength indication(RSSI). Using three antennas to obtain channel state information of WiFi to estimate yaw angle of arrive(AOA) has become reality. Based on orthogonal frequency division multiplexing (OFDM) technology, this paper uses two antennas instead of three antennas to create a virtual antenna array with 60 antennas instead of 90 antennas and extends the forward smoothing algorithm to the two-dimensional forward smoothing algorithm, then uses experimental data of non-coherent signal and coherent signal in view of multiple signal classification(MUSIC) to verify the algorithm proposed, which can realize yaw angle of arrive estimation just with two antennas instead of three antennas. The virtual antenna array model and two-dimensional forward smoothing algorithm with two antennas proposed in this paper have validity and applicability.