Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Non-Linear Precoding via Dirty Paper Coding for Near-Field Downlink MISO Communications

Akash Kulkarni, Rajshekhar V Bhat

TL;DR

A nonlinear precoding framework based on Dirty Paper Coding is proposed, which pre-cancels known interference to maximize the sum-rate performance and achieves substantial sum-rate gains over ZF across various near-field configurations.

Abstract

In 6G systems, extremely large-scale antenna arrays operating at terahertz frequencies extend the near-field region to typical user distances from the base station, enabling near-field communication (NFC) with fine spatial resolution through beamfocusing. Existing multiuser NFC systems predominantly employ linear precoding techniques such as zero-forcing (ZF), which suffer from performance degradation due to the high transmit power required to suppress interference. This paper proposes a nonlinear precoding framework based on Dirty Paper Coding (DPC), which pre-cancels known interference to maximize the sum-rate performance. We formulate and solve the corresponding sum-rate maximization problems, deriving optimal power allocation strategies for both DPC and ZF schemes. Extensive simulations demonstrate that DPC achieves substantial sum-rate gains over ZF across various near-field configurations, with the most pronounced improvements observed for closely spaced users.

Non-Linear Precoding via Dirty Paper Coding for Near-Field Downlink MISO Communications

TL;DR

A nonlinear precoding framework based on Dirty Paper Coding is proposed, which pre-cancels known interference to maximize the sum-rate performance and achieves substantial sum-rate gains over ZF across various near-field configurations.

Abstract

In 6G systems, extremely large-scale antenna arrays operating at terahertz frequencies extend the near-field region to typical user distances from the base station, enabling near-field communication (NFC) with fine spatial resolution through beamfocusing. Existing multiuser NFC systems predominantly employ linear precoding techniques such as zero-forcing (ZF), which suffer from performance degradation due to the high transmit power required to suppress interference. This paper proposes a nonlinear precoding framework based on Dirty Paper Coding (DPC), which pre-cancels known interference to maximize the sum-rate performance. We formulate and solve the corresponding sum-rate maximization problems, deriving optimal power allocation strategies for both DPC and ZF schemes. Extensive simulations demonstrate that DPC achieves substantial sum-rate gains over ZF across various near-field configurations, with the most pronounced improvements observed for closely spaced users.
Paper Structure (15 sections, 1 theorem, 40 equations, 3 figures)

This paper contains 15 sections, 1 theorem, 40 equations, 3 figures.

Table of Contents

  1. Introduction
  2. Literature Survey
  3. Contributions
  4. System Model and Problem Formulation
  5. Setup and Near-Field Channel Model
  6. Signal Model and Precoding Framework
  7. Problem Formulation
  8. Zero-Forcing: Linear, Post-Nulling Interference Cancellation
  9. Encoding Order Selection and Channel Decomposition
  10. DPC Precoding Structure
  11. Successive Encoding with Interference Pre-Cancellation
  12. Sum-Rate Maximisation with DPC
  13. Analytical Characterisation of Precoding Performance
  14. Numerical Results
  15. Conclusion

Key Result

Theorem 1

Consider a two-user co-linear near-field MISO system with a uniform planar array (UPA) of $N = N_x \times N_y$ antennas centred at the origin on the $xy$-plane. Two users are positioned along the $z$-axis at $\mathbf{r}_1 = [0, 0, d]^T$ and $\mathbf{r}_2 = [0, 0, d + \Delta]^T$, where $d > 0$ is fix

Figures (3)

  • Figure 1: System model for a generic MISO system with K users.
  • Figure 2: Achievable rate regions comparing DPC and ZF. (a) Co-linear: users aligned radially. (b) Coplanar: users separated angularly. DPC achieves $147\%$ larger rate region area than ZF in co-linear case. The parameters chosen are: $N_x= 500$, $d=10$, $P_t =10$, $\Delta =0.2$.
  • Figure 3: (a) Variation of $\alpha_k$ and $(r_{kk}^{(\pi^\star)})^2$ with inter-user spacing $\Delta$. (b) Sum-rate trend for $K$ users arranged colinearly along the $z$-axis within a range of $2$ units. (c), (d) Sum-rate gain comparisons for various distances $d$ and spacings $\Delta$ with $N_x = 10$ and $1000$, respectively.

Theorems & Definitions (2)

  • Theorem 1
  • proof