High data rates are one of the most important requirements of wireless telecommunication users. New generation telecommunication systems have been able to achieve this goal by using several technologies. Combining massive MIMO and high-performance relaying capabilities has greatly helpedtomeetthisneed, byincreasingthespectralefficiency, aswellas, theexpansionofthecoverage of wireless networks and maintaining capacity even for cell’s edge users. On the other hand, due to the high number of antennas at the massive MIMO base stations and multi-antenna relays, the practical implementation of these systems confronts challenges. These include physical constraints, hardware and computational complexities and, high number of RF circuits leading to high costs. The idea of antenna selection, proposes the activation of a subset of antennas with the best transmit conditions, based on a logical selection criterion, for the data, while maintaining the benefits of these systems alongside circumventing the challenges. The use of linear processing such as ZF in massive MIMO systems is important because of their near optimal performance. In a recent work, the downlink transmit sum rate of a massive MIMO relay system using linear precoding has been investigated. If a base station haMantennas and relay harantennas in order to serveKusers, the precoding matrix at the base station will be aM×Nrmatrix. Therefore,onlyKcolumns of the precoding matrix are required. In this thesis, we have attempted to improve the system sum rate by iiring from a common norm-based antenna selection algorithm. The selection of appropriate subsets of the precoding matrix and antenna selection in the relay prior to the precoding design are two proposed methods in this thesis. The third proposed method is to select antennas in the transmitter of the base station and the relay concurrently (by channel norm), which will reduce the required number of RF circuits and the hardware complexity of the system. The results show that applying any of the proposed methods can improve the system sum rate compared to the random selection of columns from the ZF precoding matrix. Also, it is observed 7? that the performance of the second proposed method is better than the first one, despite its complexity. The third approach, shows excellent performance, but imposes high computational complexity. KeyWords : Massive MIMO Relaying Systems, Norm Based Antenna Selection, ZF Precoding, SumRate