[Paper Review] A Fair Individual Rate Comparison between MIMO-NOMA and MIMO-OMA

Ming Zeng, Animesh Yadav+2 more|arXiv (Cornell University)|Sep 15, 2017

Advanced Wireless Communication Technologies3 citations

TL;DR

This paper proposes a power allocation (PA) strategy for MIMO-NOMA that guarantees higher individual rates for all users compared to MIMO-OMA, even when OMA uses arbitrary power allocation and optimal degrees of freedom (DoF) splitting. The key contribution is a theoretically proven PA scheme ensuring individual rate superiority and improved fairness, validated through analytical derivations and simulations under realistic MIMO downlink conditions with user pairing and beamforming.

ABSTRACT

In this paper, we compare the individual rate of MIMO-NOMA and MIMO-OMA when users are paired into clusters. A power allocation (PA) strategy is proposed, which ensures that MIMO-NOMA achieves a higher individual rate for each user than MIMO-OMA with arbitrary PA and optimal degrees of freedom split. In addition, a special case with equal degrees of freedom and arbitrary PA for OMA is considered, for which the individual rate superiority of NOMA still holds. Moreover, it is shown that NOMA can attain better fairness through appropriate PA. Finally, simulations are carried out, which validate the developed analytical results.

Motivation & Objective

• To address the lack of fair individual rate comparison between MIMO-NOMA and MIMO-OMA in existing literature.
• To develop a power allocation strategy ensuring MIMO-NOMA achieves higher individual rates than MIMO-OMA under arbitrary PA and optimal DoF splitting.
• To demonstrate that NOMA maintains individual rate superiority even when DoF are equally split between users.
• To show that NOMA can achieve better fairness through appropriate power allocation.
• To validate analytical results through simulations under realistic MIMO downlink system models with user pairing and beamforming.

Proposed method

• Proposes a novel power allocation strategy for MIMO-NOMA that ensures individual rates exceed those in MIMO-OMA under arbitrary OMA power allocation and optimal DoF splitting.
• Derives a closed-form PA solution using signal-to-interference-plus-noise ratio (SINR) constraints and beamforming alignment conditions.
• Applies zero-forcing (ZF) precoding at the base station to suppress inter-cluster interference.
• Imposes signal alignment at the receiver between users in the same cluster via receive beamforming vectors.
• Uses a block fading channel model with path loss and Rayleigh fading components for realistic channel representation.
• Validates analytical results through simulations with M=4 antennas, ρ=30 dB, and varying power coefficients.

Experimental results

Research questions

• RQ1 Does MIMO-NOMA achieve higher individual rates than MIMO-OMA when OMA uses arbitrary power allocation and optimal DoF splitting?
• RQ2 Can MIMO-NOMA maintain individual rate superiority when DoF are equally split between users and OMA uses arbitrary power allocation?
• RQ3 How does power allocation in MIMO-NOMA affect fairness compared to MIMO-OMA?
• RQ4 What is the impact of optimal DoF splitting on the sum rate and individual rate comparison between NOMA and OMA?
• RQ5 Can the proposed PA strategy ensure individual rate gains across all user pairs under realistic MIMO downlink conditions?

Key findings

• The proposed PA strategy ensures that MIMO-NOMA achieves strictly higher individual rates than MIMO-OMA for all users, even when OMA uses arbitrary power allocation and optimal DoF splitting.
• For the special case of equal DoF and arbitrary PA, MIMO-NOMA still achieves higher individual rates than MIMO-OMA, as analytically proven.
• Simulations confirm that NOMA outperforms OMA in individual rate across all user pairs, with both $R_1$ and $R_2$ in NOMA exceeding their OMA counterparts when optimal PA is applied.
• Under NOMA with PA following equation (12b), both $R_1$ and $R_2$ are higher than in OMA [15], which assumes equal power and DoF, demonstrating NOMA's superiority even against suboptimal OMA schemes.
• The sum rate order is $\text{NOMA}_3 > \text{NOMA}_4 > \text{OMA} > \text{OMA}_{[15]}$, confirming that OMA with optimal DoF achieves higher sum rate than equal-DoF OMA.
• Figure 1 shows that NOMA can achieve better fairness than OMA when the strong user's power coefficient is low, as $R_1^{\text{NOMA}_1} = R_1^{\text{OMA}} > R_2^{\text{NOMA}_1} > R_2^{\text{OMA}}$ for $\alpha_{2'}^2 \in [0, 0.8]$.