In this paper, we investigate the issue of self-interference mitigation in multiple input multiple output (MIMO) channel. Indeed, several macro base stations (MBSs) network transmit simultaneously signals to users of their own small cells, which causes interference. Hence, the main issue consists of designing a linear beamformers allowing the maximization of system throughput. A promising method for designing beamformers is the weighted mean square error (WMSE) for maximization sum rate (SR). By adopting this approach, we design a beamformer that can significantly maximizes the weighted SR to per base station power constraint. The presented simulation results show that our proposal converges to a fixed point of the weighted SR maximization objective function.