Reciprocal Training and Scheduling Protocol for MIMO systems

We propose a new training based scheme for multiple antenna broadcast channels consisting of a base station (transmitter) and many users (receivers) in a flat fading environment. Both the transmitter and the users transmit at a common carrier frequency so that, by virtue of reciprocity, the downlink and uplink propagation matrices are equal. We show that by using appropriate training the base station and users to learn the channel state information efficiently. With the knowledge of the channel at both ends of the link, they diagonalize the channel optimally and schedule power settings in order to maximize a performance criterion. This permits independent coding over resulting sub-channels of the diagonalized channel. The proposed training scheme also automatically allows both ends of the link to agree on the rate scheduling. The practical advantages of this scheme are (a) the simplicity of coding and (b) its robustness to the nature of the flat fading environment. In particular, the scheme provides a seamless transition between Rayleigh propagation to specular (beamforming) propagation.