Feedback-assisted multi-antenna transmission weight adaptation for wireless communications

This dissertation introduces and studies a class of feedback assisted stochastic gradient based algorithms for the adaptation of multi-antenna complex baseband transmission weights in a wireless communication system. The system benefits by adapting those weights such that the radio waves from each transmit antenna add constructively at the receive antenna(s), so that a given received signal power can be attained with a smaller transmit power. This can provide performance gains from both the increased received power and from diversity in a fading environment. Versions of the algorithm are introduced for both “multiple-input single-output” and “multiple-input multiple-output” antenna systems.; The algorithms which are introduced utilize a random perturbation probing strategy with feedback from a receiving device which indicates the preferred perturbation update direction of the weights. This provides the transmitting device with an estimate of the gradient of a received power metric with respect to the weights. The statistics of this perturbation update strategy are derived.; A general formulation of the algorithm for multiple-input single-output systems is introduced. Performance tradeoffs for dedicated or common perturbation probe transmission are introduced and considered, as well as performance enhancements in the presence of spatial fading correlation of the transmit antenna array to receive antenna channel gain vector. The convergence behavior and dynamic tracking performance of the algorithm is analyzed, with comparisons of various realizations of the algorithm. Monte Carlo computer simulation confirms the analyses, as well as providing a comparison to previously proposed diversity space time coded and codebook weight selection feedback systems.; Subspace tracking for transmission adaptation in multiple-input multiple-output systems is introduced, with a formulation of the algorithm proposed as a specific means to accomplish such tracking. The multiple spatial channels of the multiple-input multiple-output system provide increased effective communications bandwidth, in the form of increased orthogonal-bases-per-second, which increases the achievable bit rate. In this context the adaptation algorithm provides increased performance and reliability. The behavior and performance of the algorithm is explored through analysis and Monte Carlo computer simulations.