Performance of precoded integer-forcing for closed-loop MIMO multicast

The integer-forcing receiver architecture has recently been proposed as a high-performance, yet low-complexity, equalization scheme, that is applicable when all data streams are encoded with the same linear code. It was further shown in [1], that this receiver architecture, when coupled with space-time linear precoding is able to achieve the capacity of the open-loop multiple-input multiple-output channel, up to a constant gap that depends only on the number of transmit antennas. The gap, however, is quite large and thus provides performance guarantees that are useful only for high values of capacity. In this work, we consider the problem of multicast over multiple-input multiple-output channels to a modest number of users, and with space-only linear precoding. It is assumed that channel state information is available to the transmitter, allowing it to optimize the precoding matrix so as to maximize the achievable transmission rate. It is numerically demonstrated that this architecture allows to very closely approach the multicast capacity at all transmission rates regimes.