Abstract : We present a new precoding model for closed-loop multiple-antenna systems based on a tensor signal modeling and transmit antenna selection. The precoder is modeled as a third-order tensor and decomposed as a function of (i) an antenna selection matrix determining the allocation of data streams to transmit antennas and (ii) an orthogonal code matrix that spreads the data streams across time-slots. A sub-optimum transmit antenna selection algorithm based on limited-feedback is proposed. Blind channel estimation and symbol detection is afforded at the receiver by exploiting the algebraic tensor structure of the received signal. The bit error rate of the closed-loop system is tested by means of computer simulations.