Quality of service assurance in multi-antenna relay-assisted networks

This study concerns two-hop communication over a relay-assisted wireless network, including L point-to-point communication links and a multi-antenna relaying node. It is assumed the relay is equipped with N_r antennas and the amplify-and-forward strategy is employed at this node, considering the second-order statistics of channel coefficients are merely available at the relay. To this end, each source sends its desired signal to the relay during the first hop. Thus, the relay simultaneously receives a faded version of L transmitted signals corrupted with an additive white Gaussian noise in a vector form. Then, the relay applies a weight matrix of dimension N_r×N_r to the received vector and transmits the resulting vector to the destinations throughout the second hop. In this regard, the task of finding the best weight matrix at the relaying node under two different criteria, that is, minimising the relay´s transmit power to have a minimum signal-to-interference noise ratio (SINR) at destinations and maximising the worst-case SINR for a given relay´s transmit power, is investigated. Accordingly, it is demonstrated that the corresponding problems can be formulated as optimisation problems which are not convex in general. This motivated the authors to propose suboptimal solutions through using the so-called semidefinite relaxation method to translate original problems into semidefinite programming problems. Numerical results are provided, showing the superiority of the proposed methods as compared to the best known method.