In this paper, we propose a linear precoding and decoding scheme maximizing the sum rate of K-user interference channel systems where each node has multiple antennas. With an iterative approach, the precoding matrices are identified by deriving the gradient of the sum rate and applying the gradient descent method. Due to non-convexity of the formulated problem, the proposed precoder cannot guarantee the global optimal solution, and a locally maximized sum rate can be found by the proposed precoding scheme. Then, we obtain the single-symbol decodable receiver from the modified minimum mean-squared error filter. From simulation results, we exhibit a local optimal sum rate of the interference channel systems with the proposed method. Also, we demonstrate that the proposed algorithm outperforms other existing methods in terms of the sum rate.