Implementation of Approximate Matrix Inversion for Multi – User Cellular Systems Kalyani K*, Rajaram S** *Assistant Professor, Department of ECE, Thiagarajar College of Engineering, Madurai, Tamilnadu, India **Associate Professor, Department of ECE, Thiagarajar College of Engineering, Madurai, Tamilnadu, India Online published on 9 November, 2016. Abstract Large scale MIMO is a prominent technology for the next generation 5G multi -user cellular systems which offers high data rate with better reliability and less power consumption. In this work, we propose VLSI architecture for an approximate matrix inversion algorithm relying on Neumann series expansion for linear data detection in massive MIMO systems. Then VLSI architecture based on systolic arrays for the approximate inverse based on Neumann series expansion and exact matrix inversion algorithms are simulated, synthesized using Xilinx 14.1 tool using VHDL. Reference implementation is done on Xilinx Virtex 7 XC7VH290T device. Finally, resource utilization for the proposed matrix inversion algorithm and conventional matrix inversion algorithm is provided to illustrate the hardware complexity. Implementation results reveal that for the large scale MIMO systems with the count of 64 base station antennas and 4 users, the proposed matrix inversion algorithm is able to reduce the hardware implementation complexity considerably than the conventional matrix inversion algorithm. Top Keywords Cellular systems, VLSI, Neumann series expansion, Large scale MIMO, Data detection. Top |