Currently, most antenna systems are built based on MIMO technology. Its main advantage compared to classical antenna arrays is the possibility of a significant reduction of the number of physical transmitting and receiving channels keeping the resolution in angular coordinates. This feature is achieved due to the fact that the transmitting or receiving array is designed to be sparse, that is, the distance between the phase centers is significantly higher than half the wavelength. The cost of simplifying the design is presence of energy losses. In addition, the presence of several transmitting signals leads to the need to separate them from each other at the receiving position. The most popular method for separating orthogonal signals in MIMO arrays is code division signal separation. To implement this method, various types of orthogonal sequences are used. The most common are the Walsh, Fourier, and pseudo-random sequences. The use of orthogonal codes leads to the arising of multiple maximums in two-dimensional spectrum of the signal reflected from the object, exactly in dimension of Doppler frequency, with regular intervals. The number of maximums corresponds to the number of orthogonal sequences. Moreover, these maximums are formed in both case when using a sequence decoder and in the absence of a decoder. This feature is explained by the fact that when the sum of signals modulated in phase by Walsh sequences is multiplied by one of these sequences, one of the signals is demodulated, and the rest receive additional phase modulation. The level of additional maximum corresponds to the level of the main maximum. As a result, the ambiguity in determining the Doppler frequency appears. The indicated problem is very actual at the present time. To solve this problem, in the present paper the method of spatial-temporal coding of signals in millimeter-wave antenna array has been considered, based on the simultaneous operation of all sparse transmitters encoded by Walsh or Fourier sequences and spatial selection of transmitting beams by receiving one. The geometry of the placement of the elements of the transmitting and receiving antenna arrays corresponds to the geometry of the elements of MIMO array. At the same time, in contrast to the classical approach of beam forming for MIMO array, when individual transmitting antennas radiate orthogonal signals, and the irradiation of the target by each of them is considered separately, in the proposed method, the target is irradiated with a beam formed for transmission due to coherent spatial addition of signals from individual sparse transmitting array elements shifted in phase in accordance with the required slope of the phase front. As a result, the directivity pattern (DP) of the transmitting and receiving antenna is formed, which is equal to the product of the DP of the transmitting and receiving arrays. In this case, the interference maxima of DP of the transmitting antenna are at the zeros of DP of the receiving antenna, and the resulting DP coincides with the DP of the virtual receiving antenna formed using the classical MIMO approach. In the course of the studies and simulations, results have been obtained that show the effectiveness of the described method in terms of Doppler frequency ambiguity elimination when encoding signals in MIMO antenna arrays of the millimeter range.