多天線非正交多工接取之公平排程技術設計與研究

Proportional Fair Scheduling for Non-Orthogonal Multiple Access in Multiple Antenna Systems

胡靜文(Ching-Wen Hu)

指導教授 ： 謝宏昀

國立臺灣大學/電機資訊學院/電信工程學研究所/碩士(2015年)

https://doi.org/10.6342/NTU.2015.02094

摘要

在本篇論文中，我們實踐多天線非正交多工接取之核心概念，在發射端波束重疊多個用戶的信號，並在用戶端使用連續干擾消除的技巧解開信號，在假設傳送端完美的知道整個通道狀況，探討此系統搭配迫零預編碼的表現。由於傳統理想的數學模型計算可達到的資料傳輸量無法準確的反應出連續干擾消除的迭代解碼過程中發生的問題，例如內嵌的誤差傳播，我們建構實體層環境，模擬不同的調變與編碼方式所對應的位元錯誤率，藉此計算出實際可達到的資料傳輸量。
因此，在資源分配上，關鍵的四個問題包括預編碼的設定使一個波束可以服務多個重疊的使用者、波束間與波束內使用者的能量控制、調變與編碼的設定和排程的選定。
同時考慮公平性與整體的傳輸量下，採用比例公平調度的模型作為最佳化的目標函數。我們首先提出基於交叉熵的觀念建立的啟發式排程運算解決優化問題，接著，我們將問題轉換成迭代調度最佳化問題，以滿足蜂巢式網路系統排程架構。我們進一步將用戶根據不同的標準作適當的分群，例如通道的相關性和信噪比差值，使系統在不降低過多的效率下能降低運算的時間作及時的分配。最後，我們亦研究如何將非正交結合多天線之多路接取的系統套入至現今的行動網路例如LTE/LTE-A中。模擬的結果顯示，提出的架構不論是在公平性或是整體的資料傳輸量上，都遠勝於傳統的正交多天線方式約$20\%$至$30\%$，又較複雜的交叉熵排程相較於迭代方式約有$40\%$的增益。當我們適度地將用戶分群，例如使用者的通道相關性高於一定值才有機會重疊在同一波束中，使系統在複雜度減半的情況下，幾乎不損失效率。而當用戶們的平均信噪比高時，提升的效益甚至超越$50\%$。因此，同時利用空間域與能量域資源的多天線非正交多工接取，被視為未來極具發展性的技術。

關鍵字

多天線 ； 非正交 ； 公平排程

並列摘要

In this thesis, we investigate non-orthogonal multiple access (NOMA) combined with multiple-input multiple-output (MIMO) system with zero-forcing (ZF) beamforming with the assumption of perfect CSIT. The key concept of the NOMA-MIMO is to superpose multiple user signals within a beam at the transmitter and apply successive interference cancellation (SIC) at the user terminal. Rather than using the ideal mathematical capacity model, we construct the physical-layer adaptive modulation and coding scheme (AMC) to accurately reflect the error propagation effect during the iterative decoding process.
Therefore, the critical design issues on NOMA-MIMO include the precoding setting, the transmission power control among/within beams, the MCS selection and the scheduling strategy. In terms of fairness and sum rate, the proportional fair (PF) scheduling model is adopted to evaluate system performance. We first propose the metaheuristic scheduling approach based on the cross entropy method to solve the optimization problem of maximizing product of user sum rates.
Then, we relax the maximization problem to the iterative scheduling scheme which is applied to the realistic systems such as LTE/LTE-A.
To further reduce the computational complexity but not lose performance much, we consider to divide users into groups based on various criteria.
In addition, the implementation of the NOMA-MIMO scenario in nowadays communication network is also taken into account.
Simulation results show the proposed NOMA-MIMO system significantly outperforms conventional OMA-MIMO in not only system sum rate but also fairness by around $20\%$ to $30\%$. The more complicated cross entropy method improves $40\%$ gains compared with the iterative scheme. If we appropriately group users such as intra-beam correlation threshold, the perforance is almost the same while the complexity reduces by half. In addition, as users' average SINRs are high, NOMA-MIMO even achieve more than $50\%$ gain. Therefore, NOMA-MIMO is expected to be a promising technique in the next generation of the cellular mobile communication which provides superior spectral efficiency by exploiting not only spatial-domain but also power-domain resources.

並列關鍵字

MIMO ； NOMA ； PF

參考文獻

[27] Spatial Channel Model for Multiple Input Multiple Output (MIMO) Simula-

and candidate technologies for future steps of LTE-A,” in Communication

[2] Y. Saito and Y. Kishiyama, “Non-orthogonal multiple access (NOMA) for cel-

2013 IEEE 77th, June 2013, pp. 1–5.

[6] S. Timotheou, “Fairness for non-orthogonal multiple access in 5G systems,”