基于提前終止迭代的概率近似消息傳遞檢測(cè)算法

申敏; 任茜源; 何云

doi:10.11999/JEIT190471

基于提前終止迭代的概率近似消息傳遞檢測(cè)算法

doi: 10.11999/JEIT190471 cstr: 32379.14.JEIT190471

1.
重慶郵電大學(xué)通信與信息工程學(xué)院重慶 400065
2.
重慶郵電大學(xué)通信核心芯片、協(xié)議及系統(tǒng)應(yīng)用團(tuán)隊(duì) 重慶 400065

基金項(xiàng)目: 國(guó)家科技重大專項(xiàng)基金(2018ZX03001026-002)

詳細(xì)信息

作者簡(jiǎn)介:
申敏：女，1963年生，教授，研究方向?yàn)橥ㄐ藕诵男酒?、協(xié)議與系統(tǒng)應(yīng)用技術(shù)

任茜源：女，1995年生，碩士生，研究方向?yàn)橐苿?dòng)通信物理層算法、信號(hào)檢測(cè)

何云：女，1979年生，博士生，研究方向?yàn)橐苿?dòng)通信物理層算法、混合預(yù)編碼

通訊作者:
任茜源　18883259691@163.com

中圖分類號(hào): TN929.5
計(jì)量
- 文章訪問數(shù): 1350
- HTML全文瀏覽量: 848
- PDF下載量: 65
- 被引次數(shù): 0
出版歷程
- 收稿日期: 2019-06-25
- 修回日期: 2020-04-21
- 網(wǎng)絡(luò)出版日期: 2020-08-29
- 刊出日期: 2020-11-16

Probability Approximation Message Passing Detection Algorithm Based on Early Termination of Iteration

1.
Communication and Information Engineering, Chongqing University of Posts and Telecommunications, Chongqing 400065, China
2.
Innovation Team of Communication Core Chip, Protocols and System Application, Chongqing University of Posts and Telecommunications, Chongqing 400065, China

Funds: The National Science and Technology Major Project of China (2018ZX03001026-002)

摘要

摘要: 大規(guī)模多輸入多輸出技術(shù)作為第5代通信系統(tǒng)的關(guān)鍵技術(shù)，可有效提高頻譜利用率?；径瞬捎孟鬟f檢測(cè)(MPD)算法可以實(shí)現(xiàn)良好的檢測(cè)性能。但是由于MPD算法的計(jì)算復(fù)雜度隨調(diào)制階數(shù)和用戶天線數(shù)的增加而增加，而概率近似消息傳遞檢測(cè)(PA-MPD)算法可以減少M(fèi)PD算法的計(jì)算復(fù)雜度。為了進(jìn)一步降低PA-MPD算法的復(fù)雜度，該文在PA-MPD算法的基礎(chǔ)上引入了提前終止迭代策略，提出了一種改進(jìn)的概率近似消息傳遞檢測(cè)算法(IPA-MPD)。首先確定不同用戶的符號(hào)概率在迭代過程中的收斂速率，然后根據(jù)收斂率來判斷用戶的符號(hào)概率是否達(dá)到最佳收斂，最后對(duì)符號(hào)概率到達(dá)最佳收斂的用戶終止算法迭代。仿真結(jié)果表明，在不同單天線用戶配置下IPA-MPD算法的計(jì)算復(fù)雜度可降低為PA-MPD算法的52%～77%，且不損失算法的檢測(cè)性能。
- 大規(guī)模MIMO /
- 消息傳遞檢測(cè) /
- 概率近似消息傳遞檢測(cè) /
- 提前終止迭代
Abstract: As a key technology of the fifth generation communication system, large-scale Multi-Input and Multi-Output(MIMO) technology can effectively improve spectrum utilization. The base station side uses the Message Passing Detection (MPD) algorithm to achieve good detection performance. However, the computational complexity of the MPD algorithm increases with the increase of the modulation order and the number of user antennas, and the Probability Approximation Message Passing Detection (PA-MPD) algorithm can reduce the computational complexity of the MPD algorithm. In order to further reduce the complexity of PA-MPD algorithm, this paper introduces an early termination iteration strategy based on PA-MPD algorithm, and proposes an Improved PA-MPD (IPA-MPD) algorithm. Firstly, the convergence rate of the symbol probability of different users in the iterative process is determined, and then the convergence probability is used to determine whether the user’s symbol probability reaches the best convergence. Finally, the user termination algorithm that the symbol probability reaches the best convergence is iterated. The simulation results show that the computational complexity of the IPA-MPD algorithm can be reduced to 52%～77% of the PA-MPD algorithm under different single-antenna user configurations without loss of the detection performance of the algorithm.

HTML全文

圖 1 MPD算法的消息傳遞過程

下載: 全尺寸圖片幻燈片

圖 2 不同閾值下IPA-MPD算法性能對(duì)比

下載: 全尺寸圖片幻燈片

圖 3 不同調(diào)制方式下IPA-MPD算法與PA-MPD算法性能對(duì)比

下載: 全尺寸圖片幻燈片

圖 4 3種調(diào)制方式下兩種算法的性能對(duì)比

下載: 全尺寸圖片幻燈片

圖 5 不同閾值下IPA-MPD算法計(jì)算復(fù)雜度對(duì)比

下載: 全尺寸圖片幻燈片

圖 6 不同調(diào)制方式下IPA-MPD算法與PA-MPD算法計(jì)算復(fù)雜度對(duì)比

下載: 全尺寸圖片幻燈片

圖 7 3種調(diào)制方式下IPA-MPD與PA-MPD計(jì)算復(fù)雜度比值曲線

下載: 全尺寸圖片幻燈片

表 1 IPA-MPD算法

　輸入：$J,Z,\sigma _v^2,\varDelta ,T$

　輸出：L

　1: 初始化：${p_i}({s_k}) = \dfrac{1}{ {\sqrt M } },i = 1,2, ··· ,2K,k = 1,2, ··· ,\sqrt M ,$
　　${R^1}({x_j}) = 1 $

　2: ${\rm{for} }\;t = 1\;{\rm{do}}$

　3: 　　${\rm{for} }\;i = 1\;{\rm{to}}\;2K\;{\rm{do}}$
　4: 　　　$ {{\mu }_{i}}\leftarrow \displaystyle\sum\limits_{j=1,j\ne i}^{2K}{{{J}_{ij}}\displaystyle\sum\limits_{\forall s\in \mathbb{B}}{sp_{j}^{t-1}(s)}} $

　5: 　　　$\sigma _{i}^{2}\leftarrow \displaystyle\sum\limits_{j=1,j\ne i}^{2K}J_{ij}^{2}\left(\displaystyle\sum\limits_{\forall s\in \mathbb{B} }{ { {s}^{2} }p_{j}^{t-1}(s)}-E{ {({ {x}_{j} })}^{2} } \right)+\sigma _{v}^{2}$

　6: 　　　$ {{L}_{i}}\leftarrow \dfrac{2{{J}_{ii}}}{\sigma _{i}^{2}}({{z}_{i}}-{{\mu }_{i}}) $

　7: 　　　${ { {\tilde{p} } }_{i} }\leftarrow \dfrac{ { {{\rm{e}}}^{ { {L}_{i} } } }}{1+{ {{\rm{e}}}^{ { {L}_{i} } } }}$

　8: 　　　${p_i} \leftarrow (1 - \varDelta ){ {\tilde p}_i} + \Delta { {\tilde p}_i}$

　9: 　　　$ {A_i} \leftarrow {\rm{sort}}({\rm{ }}{p_i}) $

　10: 　　end

　11: end

　12: ${\rm{for} }\;t = 2\;{\rm{to}}\;{T_{\max } }\;{\rm{do}}$

　13: 　　${\rm{for} }\;i = 1\;{\rm{to}}\;2K\;{\rm{do}}$

　14: 　　　$ {\rm{if}}\;{R^{t - 1}}({x_i}) < T$

　15: 　　　　終止迭代

　16: 　　　else
　17: 　　　　${\mu _i} \leftarrow \displaystyle\sum\limits_{j = 1,j \ne i}^{2K} { {J_{ij} }\displaystyle\sum\limits_{p_j^{t - 1}(s) \in {A_j}(1,2, ··· ,M)} {sp_j^{t - 1}(s)} }$

　18: 　　　　$\sigma _i^2 \leftarrow \displaystyle\sum\limits_{j = 1,j \ne i}^{2K} {J_{ij}^2}\left (\displaystyle\sum\limits_{p_j^{t - 1}(s) \in {A_j}(1,2, ··· ,M)} {sp_j^{t - 1}(s)} - \right.$
　　　　　　$ \Biggl.{19} E{({x_j})^2} \Bigggr){19}+ \sigma _v^2 $

　19: 　　　　$ {L_i} \leftarrow \dfrac{{2{J_{ii}}}}{{\sigma _i^2}}({z_i} - {\mu _i}) $

　20: 　　　　${ {\tilde p}_i} \leftarrow \dfrac{ { {{\rm{e}}^{ {L_i} } } }}{ {1 + {{\rm{e}}^{ {L_i} } } }}$

　21: 　　　　${p_i} \leftarrow (1 - \varDelta ){ {\tilde p}_i} + \Delta { {\tilde p}_i}$

　22: 　　　　$ {A_i} \leftarrow {\rm{sort}}({\rm{ }}{p_i}) $

　23: 　　　　$ {R^t}({x_i}) \leftarrow \displaystyle\sum\limits_{k = 1}^{\sqrt M } {|p_{{x_i}}^t({s_k}) - p_{{x_i}}^{t - 1}({s_k})|} $

　24: 　　　end

　21: 　　end

　22: end

下載: 導(dǎo)出CSV

表 2 M-QAM調(diào)制下PA-MPD^[10]算法和IPA-MPD算法的實(shí)數(shù)域乘法和加法次數(shù)

算法名稱	加法	乘法
PA-MPD-n	$\begin{array}{l}((2n + 1)(2K - 1) - 2)2K(t - 1)\\ + ((2\sqrt M {\rm{ + 1}})2K - 1 - {\rm{1}}){\rm{2}}K\end{array}$	$\begin{array}{l} (2n + 1)(2K - 1)2K(t - 1) \\ + (2\sqrt M + 1)(2K - 1)2K \\ \end{array} $
IPA-MPD-n	$\begin{array}{l}((2n + 1)(2K - 1) - 2)2K({t_{{\rm{ave}}}} - 1)\\ + ((2\sqrt M {\rm{ + 1}})2K - 1 - {\rm{1}}){\rm{2}}K\end{array}$	$\begin{array}{l}(2n + 1)(2K - 1)2K({t_{{\rm{ave}}}} - 1)\\ + ((2\sqrt M {\rm{ + 1}})2K - 1 - {\rm{1}}){\rm{2}}K\end{array}$

下載: 導(dǎo)出CSV

參考文獻(xiàn)(17)

HE Hengtao, WEN Chaokai, JIN Shi, et al. A model-driven deep learning network for MIMO detection[C]. 2018 IEEE Global Conference on Signal and Information Processing, Anaheim, USA, 2018: 584–588.

DUANGSUWAN S and JAMJAREEGULGARN P. Detection of data symbol in a massive MIMO systems for 5G wireless communication[C]. 2017 International Electrical Engineering Congress, Pattaya, Thailand, 2017: 1–4.

YANG Shaoshi and HANZO L. Fifty years of MIMO detection: The road to large-scale MIMOs[J]. IEEE Communications Surveys & Tutorials, 2015, 17(4): 1941–1988. doi: 10.1109/COMST.2015.2475242

FREY B J. Graphical Models for Machine Learning and Digital Communication[M]. Cambridge: The MIT Press, 1998: 25–34.

SOM P, DATTA T, CHOCKALINGAM A, et al. Improved large-MIMO detection based on damped belief propagation[C]. 2010 IEEE Information Theory Workshop on Information Theory, Cairo, Egypt, 2010: 1–5.

USAMI T, NISHIMURA T, OHGANE T, et al. BP-based detection of spatially multiplexed 16-QAM signals in a fully massive MIMO system[C]. 2016 International Conference on Computing, Networking and Communications, Kauai, USA, 2016: 166–170.

SOM P, DATTA T, SRINIDHI N, et al. Low-complexity detection in large-dimension MIMO-ISI channels using graphical models[J]. IEEE Journal of Selected Topics in Signal Processing, 2011, 5(8): 1497–1511. doi: 10.1109/JSTSP.2011.2166950

WU Sheng, KUANG Linling, NI Zuyao, et al. Low-complexity iterative detection for large-scale multiuser MIMO-OFDM systems using approximate message passing[J]. IEEE Journal of Selected Topics in Signal Processing, 2014, 8(5): 902–915. doi: 10.1109/JSTSP.2014.2313766

NARASIMHAN T L and CHOCKALINGAM A. Channel hardening-exploiting message passing (CHEMP) receiver in large-scale MIMO systems[J]. IEEE Journal of Selected Topics in Signal Processing, 2014, 8(5): 847–860. doi: 10.1109/JSTSP.2014.2314213

ZHU Haochuan, LIN Jun, and WANG Zhongfeng. Reduced complexity message passing detection algorithm in large-scale MIMO systems[C]. The 9th International Conference on Wireless Communications and Signal Processing, Nanjing, China, 2017: 1–5.

ZENG Jing, LIN Jun, and WANG Zhongfeng. Low complexity message passing detection algorithm for large-scale MIMO systems[J]. IEEE Wireless Communications Letters, 2018, 7(5): 708–711. doi: 10.1109/LWC.2018.2813386

TAN Xiaosi, ZHONG Zhiwei, ZHANG Zaichen, et al. Low-complexity message passing MIMO detection algorithm with deep neural network[C]. Proceedings of 2018 IEEE Global Conference on Signal and Information Processing, Anaheim, USA, 2018: 559–563.

GOLDBERGER J and LESHEM A. MIMO detection for high-order QAM based on a Gaussian tree approximation[J]. IEEE Transactions on Information Theory, 2011, 57(8): 4973–4982. doi: 10.1109/TIT.2011.2159037

GU Lixin, WANG Wenjin, ZHONG Wen, et al. Message-passing detector for uplink massive MIMO systems based on energy spread transform[C]. The 27th IEEE Annual International Symposium on Personal, Indoor, and Mobile Radio Communications, Valencia, Spain, 2016: 1–6.

JIA Min, WANG Linfang, GUO Qing, et al. A low complexity detection algorithm for fixed up-link SCMA System in mission critical scenario[J]. IEEE Internet of Things Journal, 2018, 5(5): 3289–3297. doi: 10.1109/JIOT.2017.2696028

LIU Lei, YUEN C, GANG Yongliang, et al. Convergence analysis and assurance for Gaussian message passing iterative detector in massive MU-MIMO systems[J]. IEEE Transactions on Wireless Communications, 2016, 15(9): 6487–6501. doi: 10.1109/TWC.2016.2585481

YANG Chao, XU Weihong, ZHANG Zaichen, et al. A channel-blind detection for SCMA based on image processing techniques[C]. 2018 IEEE International Symposium on Circuits and Systems, Florence, Italy, 2018: 1–5.

相關(guān)文章

施引文獻(xiàn)

資源附件(0)

訪問統(tǒng)計(jì)