Adaptive transmit eigenbeamforming with stochastic unitary plane rotations in MIMO systems with linear receivers

In this work we propose two adaptive transmit eigenbeamforming techniques for closed loop MIMO systems with low feedback rate. Both algorithms track a number of the dominant eigenmodes of a fading MIMO channel. The first algorithm applies unitary transformations incrementally, while the second runs a stochastic search over the space of angles that parameterize a set of orthonormal complex vectors. The eigenmode estimate updates rely on a single feedback bit and ensure orthonormality at every instant without enforcing the condition explicitly. We introduce a cost function that minimizes both the channel induced symbol stream coupling at the receiver and the noise coloring effects of the channel equalizer. We show that our choice of cost function improves the performance when using a suboptimal linear minimum mean-square error (MMSE) receiver. Our schemes outperform the algorithm recently proposed in J. Yang and D.B. Williams (2005) (which is based on B.C. Banister and J.R. Zeidler (2003)) by 1 dB in terms of bit error probability (BEP) performance, while keeping a similar computational complexity