LTE physical layer implementation using GPU based high performance computing

Abstract

In recent years Graphics Processing Unit (GPU) has evolved as a high performance data processing technology allowing users to compute large blocks of parallel data using an array of low complexity processors. This paper proposes the implementation of compute intensive portions of 3rd Generation Partnership Project (3GPP) Long Term Evolution (LTE) physical layer using GPU. LTE employs Orthogonal Frequency Division Multiple Access (OFDMA) in downlink and Single Carrier Frequency Division Multiple Access (SC-FDMA) in uplink. Both these demand computationally complex Inverse Fast Fourier Transform (IFFT) and Fast Fourier Transform (FFT) processing at the transmitter and the receiver. The computational requirements at the base station increases significantly with the increase in number of users. This paper presents a simulation model utilizing the massively parallel architecture of GPU to reduce computation time of IFFT and FFT operations. Simulation results demonstrate that GPU provides a framework for fast data processing in this application.