I. Introduction
Wireless communication traffic has grown exponentially due to higher speed mobile applications having been launched, and this trend is expected to continue in the future [1]. To accommodate the growing traffic, the network throughput needs to be increased. Many techniques have been proposed to increase the throughput such as distributed antennas [2], three dimensional cellular systems [3] and multi-user multiple-input-multiple-output (MU-MIMO) [4]. MU-MIMO establishes concurrent communication links with several users, simultaneously, which achieves higher network throughput than single-user MIMO. Because the performance can be implemented in the downlink where a base station with many antennas transmits packets for several users with small number of the receive antennas, MU-MIMO has been recognized as a technique suitable for the massive MIMO. MU-MIMO based on the block diagonalization (BD) has been considered for the 5th generation cellular system (5G) [5]. However, though the BD provides a user with some signal streams, the number of the streams is usually set to that of the receive antennas of the user at most. Besides, even if many receive antennas are installed on the receiver of the users, the number of the streams is not always increased as far as the BD is applied, which limits not only the user throughput but also the network throughput. Overloaded MUMIMO communication systems that transmit more streams than the number of the receive antennas have been considered [6]. However, the overloaded MU-MIMO needs receiver to have complex signal detectors.