A performance comparison between linear precoding and THP on M-MIMO near-field scenarios
Abstract This paper investigates the symbol-error-rate (SER) performance of linear and nonlinear precoders in radiative near-field (NF) massive MIMO scenarios. The classical linear precoders maximum ratio transmission (MRT), zero-forcing (ZF), and minimum mean squared error (MMSE) are compared against Tomlinson–Harashima precoding (THP) variants under same-angle and spatially distributed-user configurations. Beyond a purely simulation-based comparison, the analysis clarifies the physical origin of the NF gain: spherical wavefronts reduce effective inter-user correlation by introducing depth-domain separability, which is especially relevant when users are nearly co-angular. The results show that NF propagation can provide clear SER gains over the far-field (FF) counterpart under LoS-dominant conditions and perfect CSI, while regularized nonlinear precoding, particularly MMSE-cTHP, most effectively exploits this additional degree of freedom. A Rician-factor sweep delimits the validity of this gain, which reverses in scattering-dominated channels and reaches a twofold SER reduction under strong LoS dominance ( $$\varvec{\epsilon _k = 50}$$ ϵ k = 50 dB, reducing to $$\varvec{1.28}$$ 1.28 with $$\varvec{95\%}$$ 95 % interval $$\varvec{[1.26,1.29]}$$ [ 1.26 , 1.29 ] at $$\varvec{\epsilon }_{\varvec{k}} \varvec{= 30}$$ ϵ k = 30 dB); with $$\varvec{95\%}$$ 95 % confidence intervals, the effect is shown to change sign at $$\varvec{\epsilon }_{\varvec{k}} \varvec{\approx } \varvec{20}$$ ϵ k ≈ 20 dB, below which the NF configuration is significantly worse than its FF counterpart. A CSI-error study further reveals that the advantage vanishes once the channel estimation error exceeds about $$\varvec{1\%}$$ 1 % of the channel power, so the gain is doubly conditional, on LoS dominance and on CSI accuracy, while a complexity assessment shows that the THP variants retain the same $$\varvec{\mathcal {O}}\varvec{(}\varvec{MK}^{\varvec{2}}\varvec{)}$$ O ( MK 2 ) order as ZF and MMSE, with instrumented operation counts placing their executed design cost at $$\varvec{0.67}$$ 0.67 and $$\varvec{1.32}$$ 1.32 times that of ZF/MMSE for cTHP and MMSE-cTHP, respectively. We further test whether the depth-separability mechanism can be turned into a design rule by constructing an encoding order from the
Authors
- Thiago Augusto Bruza Alves (ORCID: https://orcid.org/0000-0002-1873-1026)
- Taufik Abrão (ORCID: https://orcid.org/0000-0001-8678-2805)
- Murilo Manoel dos Reis
Institutions
- Universidade Estadual de Londrina (BR)
Publication Details
- Journal
- Annals of Telecommunications
- Published
- 2026-11-16
- DOI
- https://doi.org/10.1007/s12243-026-01239-9
- Primary Topic
- Millimeter-Wave Propagation and Modeling
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- Coordenação de Aperfeiçoamento de Pessoal de Nível Superior
- Conselho Nacional de Desenvolvimento Científico e Tecnológico
- Fundação Araucária
- Universidade Estadual de Londrina
- Conselho Nacional das Fundações Estaduais de Amparo à Pesquisa