Channel capacity for linearly precoded multiuser MIMO for discrete constellations

This paper is based on the idea of exploiting the discrete constellation alphabets in linear precoding for the downlink of multiuser (MU) MIMO. We study the effect of discrete constellation inputs on the sum rate of different linear precoders while confining ourselves to the case of two single antenna users. We show that contrary to the case of Gaussian alphabets where altruistic solutions as interference cancellation (channel inversion - CI) and interference attenuation (regularized channel inversion - RCI) are the recommended strategies for linear precoding, it is beneficial not to attenuate or cancel the MU interference, if it is coming from discrete constellations. This interference belonging to finite sized constellations has a structure which can be effectively exploited in improving the error resilience at the users. Under such a scenario, it is better to use the degrees of freedom available at the transmitter to improve the desired signal strength at the users instead of utilizing them to nullify or attenuate the undesired signals (interferences) at the users. Therefore the egoistic linear solutions as matched filter (MF) based precoding bears the potential of enhanced sum rate and improved performance as compared to the altruistic linear solutions.