Data-Driven System-Level Evaluation of MU-MIMO in Urban FDD-to-TDD Refarming

Abstract

Frequency-division duplex (FDD) to time-division duplex (TDD) spectrum refarming is often motivated by the expected gains of reciprocity-based multi-user multiple-input multiple-output (MU-MIMO) operation. However, many system-level evaluations rely on simplified user equipment (UE) placement models, which may overestimate the availability of spatially separable users in realistic urban deployments. This paper presents a measurement-driven system-level study of an urban refarming scenario, combining site-specific ray-traced propagation with UE distributions derived from minimization of drive tests (MDT) data and street-level geometry. The proposed approach is compared with uniform street-based placement and the conventional stochastic 3GPP urban macro (UMa) model under identical assumptions.The site-specific study results show that realistic UE placement reshapes how TDD MU-MIMO is exploited. The multi-user (MU) resource block (RB) share (the fraction of shared physical resource blocks) under MDT-derived placement is statistically indistinguishable from the 3GPP baseline, yet the fraction of resource blocks carrying three spatial layers drops from about 60% to 32–38%, as site-specific geometry supplies fewer jointly separable user triples. Uniform street-based placement is the least favorable case, lowering the MU RB share below both. For the considered antenna array, FDD MU-MIMO remains marginal and codebook-limited, largely insensitive to UE distribution. These findings show that UE spatial distribution is an important modeling aspect when evaluating MU-MIMO in FDD-to-TDD refarming, and that aggregate utilization alone can hide its effect.