On Implementing Sphere Decoding Algorithms in LTE-A Uplink with the Presence of CFO

To facilitate the engineering implementation of a specific detection algorithm for practical MIMO systems, it is important to know how its performance changes with certain system parameters. In this paper, we evaluate the performance of sphere decoding (SD) algorithms which fall into the category of reduced-complexity maximum likelihood (ML) algorithms, with minimum mean-square error (MMSE) algorithm used as the benchmark for comparison, in the application scenario of LTE-A uplink. The implementation scheme of SD algorithms established in this paper takes the compensation of carrier frequency offset (CFO) into consideration, and makes a replacement of MMSE with full functionalities. Based on the simulation results generated by a simulation platform compliant with 3GPP standards, the performance gain of SD with respect to MMSE is analyzed in the aspects of its varying trends related to CFO and the number of receive antennas. Consequently, the robustness of SD algorithms against CFO is established, and it is found that fewer receive antennas give rise to larger performance gain.