杜建和 花 妍 林和昀 田 沛
?
雙向MIMO中繼系統(tǒng)中一種低復(fù)雜度的聯(lián)合信道估計(jì)方法
杜建和*①花 妍①林和昀②田 沛①
①(中國(guó)傳媒大學(xué)信息工程學(xué)院 北京 100024)②(北京郵電大學(xué)信息與通信工程學(xué)院 北京 100876)
對(duì)于雙向多輸入多輸出(MIMO)中繼系統(tǒng),如何在減少中繼負(fù)擔(dān)的情況下獲得精確的信道狀態(tài)信息(CSI)成為信道估計(jì)的一個(gè)難點(diǎn)。該文針對(duì)雙向MIMO中繼系統(tǒng),提出一種低復(fù)雜度的聯(lián)合信道估計(jì)方法。所提方法在兩個(gè)用戶端同時(shí)發(fā)送正交信道訓(xùn)練信號(hào)至中繼,中繼采用所設(shè)計(jì)的放大因子放大所接收的信號(hào)并轉(zhuǎn)發(fā)至兩個(gè)用戶。每個(gè)用戶對(duì)所接收的信號(hào)構(gòu)造平行因子(PARAFAC)模型,并根據(jù)實(shí)際系統(tǒng)要求,分別設(shè)計(jì)了迭代和非迭代的兩種擬合算法對(duì)PARAFAC模型進(jìn)行擬合,從而聯(lián)合估計(jì)出所有信道的CSI。所提信道估計(jì)方法無需在中繼處進(jìn)行信道估計(jì),減輕了中繼的負(fù)擔(dān)。與已有信道估計(jì)方法相比,所提方法設(shè)計(jì)靈活,采用的擬合算法具有較低的復(fù)雜度,而且在使用較少信道訓(xùn)練信號(hào)的情況下具有較高的信道估計(jì)精度。
雙向MIMO中繼;低復(fù)雜度;信道狀態(tài)信息;平行因子
當(dāng)中繼系統(tǒng)與多輸入多輸出(Multiple-Input Multiple-Output, MIMO)技術(shù)相結(jié)合,能充分利用空間分集,進(jìn)一步提高了系統(tǒng)的性能。目前,MIMO中繼系統(tǒng)已經(jīng)引起了學(xué)術(shù)界和工業(yè)界的廣泛關(guān) 注[1,2]。大量關(guān)于MIMO中繼方面的研究工作都假設(shè)系統(tǒng)已知精確的信道狀態(tài)信息(Channel State Information, CSI)。然而在實(shí)際通信中,CSI是未知的,因此需要被估計(jì)。
針對(duì)MIMO中繼系統(tǒng),文獻(xiàn)[3]提出了一種兩階段信道估計(jì)方法,該方法能在信宿端估計(jì)出信源到中繼和中繼到信宿的信道矩陣。文獻(xiàn)[4]提出了一種基于平行因子模型(PARAllel FACtor, PARAFAC)[5,6]的信道估計(jì)方法,該方法能聯(lián)合估計(jì)出兩跳信道矩陣。文獻(xiàn)[7]對(duì)文獻(xiàn)[4]中的交替最小二乘(Alternating Least-Squares, ALS)擬合算法進(jìn)行優(yōu)化和改進(jìn),降低了計(jì)算復(fù)雜度。文獻(xiàn)[8]針對(duì)上行多用戶MIMO系統(tǒng),利用LM(Levenberg- Marquardt)擬合算法快速收斂的特性,提高了信道估計(jì)的效率。
文獻(xiàn)[3,4,7,8]都是針對(duì)于單向MIMO中繼系統(tǒng),與單向MIMO中繼系統(tǒng)相比,雙向MIMO中繼系統(tǒng)具有更高的頻譜效率[9]。然而,雙向MIMO中繼系統(tǒng)的信道估計(jì)問題也相對(duì)復(fù)雜。針對(duì)雙向MIMO中繼系統(tǒng),文獻(xiàn)[10]提出了一種級(jí)聯(lián)信道估計(jì)方法。文獻(xiàn)[11]進(jìn)一步將文獻(xiàn)[4]中的方法擴(kuò)展到了雙向MIMO中繼系統(tǒng),在用戶端估計(jì)出兩跳信道的CSI。文獻(xiàn)[12]在文獻(xiàn)[3]的基礎(chǔ)上,針對(duì)雙向MIMO中繼系統(tǒng)提出了一種新的兩階段信道估計(jì)方法。然而該方法的第1跳信道矩陣估計(jì)精度仍然依賴于第2跳信道矩陣的估計(jì)精度,而且需要兩次在中繼處對(duì)發(fā)送功率進(jìn)行優(yōu)化才能獲得較好的信道估計(jì)性能,具有較高的計(jì)算復(fù)雜度。
本文針對(duì)雙向MIMO中繼系統(tǒng),提出了一種低復(fù)雜度的聯(lián)合信道估計(jì)方法。所提方法在兩個(gè)用戶端設(shè)計(jì)相互正交的信道訓(xùn)練信號(hào)進(jìn)行發(fā)送,中繼采用不同的放大因子對(duì)接收的信號(hào)進(jìn)行放大轉(zhuǎn)發(fā),在用戶端構(gòu)造PARAFAC模型,利用迭代的P-ALS- LS(PARAFAC-ALS with linear search)和非迭代的P-KRF(PARAFAC with Khatri-Rao Factorization) 擬合算法能估計(jì)出每一跳的信道矩陣。
本文主要?jiǎng)?chuàng)新點(diǎn)如下:
(1)所提方法無需在中繼處進(jìn)行信道估計(jì),在用戶端就能估計(jì)出所有信道的CSI;而且所提信道估計(jì)方法設(shè)計(jì)靈活,可根據(jù)系統(tǒng)要求考慮信道估計(jì)精度和估計(jì)效率的折中。
(2)所提方法設(shè)計(jì)了P-ALS-LS和P-KRF算法對(duì)所構(gòu)造的PARAFAC模型進(jìn)行擬合。與文獻(xiàn)[11]提出的迭代TP-ALS (Traditional PARAFAC with ALS)算法相比,所提算法具有較低的復(fù)雜度,特別是P-KRF算法;此外,還可以根據(jù)系統(tǒng)參數(shù)來選擇合適的擬合算法。
(3)在較小的信道訓(xùn)練數(shù)目條件下,所提信道估計(jì)方法比文獻(xiàn)[11]的方法具有更高的信道估計(jì)精度。與兩階段信道估計(jì)方法[12]相比,所提信道估計(jì)方法具有更好的第2跳信道估計(jì)精度。
圖1 雙向MIMO中繼通信系統(tǒng)
由式(7)可得
其中,
根據(jù)PARAFAC模型的分解唯一性定理[6],所構(gòu)模型式(13)的唯一性條件為
為了快速而精確地估計(jì)出每一跳的信道矩陣,本文設(shè)計(jì)了兩種有效的擬合算法來擬合所構(gòu)造的PARAFAC模型,即P-ALS-LS和P-KRF擬合算法。
所提P-KRF算法的具體實(shí)現(xiàn)步驟如下:
表1 不同算法的計(jì)算復(fù)雜度比較
圖2 TP-ALS和P-ALS-LS算法達(dá)到收斂所需的迭代次數(shù)
圖3 TP-ALS, P-ALS-LS和P-KRF算法所需的平均處理時(shí)間
圖4 TP-ALS算法、所提P-KRF算法和TSCT算法的NMSE性能比較
圖5 不同系統(tǒng)參數(shù)N與下,所提算法的NMSE性能
圖6 放大矩陣和訓(xùn)練信號(hào)長(zhǎng)度對(duì)所提算法性能的影響
圖7 信道相關(guān)系數(shù)對(duì)所提算法性能的影響
本文針對(duì)雙向MIMO中繼系統(tǒng),提出了一種低復(fù)雜度的信道估計(jì)方法。所提方法能在用戶端估計(jì)出所有信道的CSI。本文詳細(xì)闡述了所提信道估計(jì)方法的建模,唯一性條件和擬合算法的設(shè)計(jì)。與已有信道估計(jì)方法相比,所提方法具有較高的信道估計(jì)精度,而且可以根據(jù)系統(tǒng)要求選擇相應(yīng)低復(fù)雜度的擬合算法,仿真驗(yàn)證了所提信道估計(jì)方法的性能。下一步的研究工作將針對(duì)雙向MIMO中繼系統(tǒng),考慮聯(lián)合信道與符號(hào)估計(jì)方案,即在假設(shè)CSI未知的條件下,無需發(fā)送信道訓(xùn)練信號(hào),僅僅利用發(fā)送的有用符號(hào)和雙向中繼信道互易性特點(diǎn),通過構(gòu)造高維的PARAFAC模型或TUCKER模型,設(shè)計(jì)相應(yīng)的擬合算法在用戶端對(duì)信道和符號(hào)進(jìn)行聯(lián)合估計(jì)。
[1] NOSRATINIA A, HUNTER T E, and HEDAYAT A. Cooperative communication in wireless networks[J]., 2004, 42(10): 74-80. doi: 10.1109 /MCOM.2004.1341264.
[2] JIN X and KIM Y H. The approximate capacity of the MIMO relay channel[J]., 2017, 63(2): 1167-1176. doi: 10.1109/TIT.2016.2632625.
[3] KONG Ting and HUA Yingbo. Optimal design of source and relay pilots for MIMO relay channel estimation[J]., 2011, 59(9): 4438-4446. doi: 10.1109/TSP.2011.2158429.
[4] RONG Yue and KHANDAKER M R A. Channel estimation of dual-hop MIMO relay systems using parallel factor analysis[C]. 2011 IEEE 17th Asia-Pacific Conference on Communications (APCC), Sabah, Malaysia, 2011: 278-283. doi: 10.1109/APCC.2011.6152819.
[5] SIDIROPOULOS N D, BRO R, and GIANNAKIS G B. Parallel factor analysis in sensor array processing[J]., 2000, 48(8): 2377-2388. doi: 10.1109/78.852018.
[6] SIDIROPOULOS N D, DE Lathauwer L, FU Xiao,. Tensor Decomposition for Signal Processing and Machine Learning[J]., 2017, 65(13): 3551-3582. doi: 10.1109/TSP.2017.2690524.
[7] DU Jianhe, YUAN Chaowei, and ZHANG Jinbo. Low complexity PARAFAC-based channel estimation for non- regenerative MIMO relay systems[J]., 2014, 8(12): 2193-2199. doi: 10.1049/iet-com.2013.1155.
[8] 林和昀, 袁超偉, 杜建和. 一種多用戶上行放大轉(zhuǎn)發(fā)中繼系統(tǒng)中快速收斂的信道估計(jì)方法[J]. 物理學(xué)報(bào), 2016, 65(21): 35-43. doi: 10.7498/aps.65.210201.
LIN Heyun, YUAN Chaowei, and DU Jianhe. A fast algorithm with convergence for channel estimation in multi- user uplink amplify-and-forward relay system[J]., 2016, 65(21): 35-43. doi: 10.7498/aps.65. 210201.
[9] SANGUINETTI L, D,AMICO A A, and RONG Yue. A tutorial on the optimization of amplify-and-forward MIMO relay systems[J]., 2012, 30(8): 1331-1346. doi: 10.1109/JSAC. 2012.120904.
[10] FANG Zhaoxi, JIONG Shi, and SHAN Hangguan. Comparison of channel estimation schemes for MIMO two-way relaying systems[C]. 2011 IEEE Cross Strait Quad-Regional Radio Science and Wireless Technology Conference (CSQRWC), 2011, 1: 719-722. doi: 10.1109/ CSQRWC.2011.6037053.
[11] RONG Yue, KHANDAKER M R A, and YONG Xiang. Channel estimation of dual-hop MIMO relay system via parallel factor analysis[J]., 2012, 11(6): 2224-2233. doi: 10.1109/TWC. 2012.032712.111251.
[12] CHIONG C W R, RONG Yue, and YONG Xiang. Channel training algorithms for two-way MIMO relay systems[J]., 2013, 61(16): 3988-3998. doi: 10.1109/TSP.2013.2265223.
[13] GHASEMI-GOOJANI S, KARIMI-BIDHENDI S, and BEHROOZI H. On the capacity region of asymmetric gaussian two-way line channel[J]., 2016, 64(9): 3669-3682. doi: 10.1109/ TCOMM.2016.2597852.
[14] 王林, 芮國(guó)勝, 張海波. 雙向中繼系統(tǒng)用戶聯(lián)合預(yù)編碼與檢測(cè)算法研究[J]. 電子與信息學(xué)報(bào), 2015, 37(7): 1550-1555. doi: 10.11999/JEIT141455.
WANG Lin, RUI Guosheng, and ZHANG Haibo. Research on users joint precoding and detection in two-way relay systems[J].&, 2015, 37(7): 1550-1555. doi: 10.11999/JEIT141455.
[15] S?RENSEN M and DE LATHAUWER L. Blind signal separation via tensor decomposition with Vandermonde factor: Canonical polyadic decomposition[J]., 2013, 61(22): 5507-5519. doi: 10.1109/TSP. 2013.2276416.
[16] FAVIER G, FERNANDES C A R, and DE ALMEIDA A L F. Nested Tucker tensor decomposition with application to MIMO relay systems using tensor space-time coding (TSTC) [J]., 2016, 128: 318-331. doi: 10.1016/ j.sigpro.2016.04.009.
[17] RAJIH M, COMON P, and HARSHMAN R A. Enhanced line search: A novel method to accelerate PARAFAC[J]., 2008, 30(3): 1128-1147. doi: 10.1137/06065577.
[18] DU Jianhe, YUAN Chaowei, HU Zhongwei,. A novel tensor-based receiver for joint symbol and channel estimation in two-hop cooperative MIMO relay systems[J]., 2015, 19(11): 1961-1964. doi: 10. 1109/LCOMM.2015.2473172.
[19] XIMENES L R, FAVIER G, and DE ALMEIDA A L F. Closed-form semi-blind receiver for MIMO relay systems using double Khatri-Rao space-time coding[J]., 2016, 23(3): 316-320. doi: 10.1109/LSP. 2016.2518699.
杜建和: 男,1984年生,博士,講師,主要研究方向?yàn)橹欣^通信、信道估計(jì)與多維信號(hào)處理技術(shù).
花 妍: 女,1988年生,博士,講師,主要研究方向?yàn)樾盘?hào)分析與處理技術(shù).
林和昀: 男,1985年生,博士生,研究方向?yàn)槎嗵炀€與信道估計(jì)技術(shù).
田 沛: 男,1970年生,博士,教授,主要研究方向?yàn)橥ㄐ排c信息網(wǎng)絡(luò)技術(shù).
A Low-complexity Algorithm for Joint Channel Estimation inTwo-way MIMO Relay Communication Systems
DU Jianhe①HUA Yan①LIN Heyun②TIAN Pei①
①(,,100024,)②(,,100876,)
For two-way Multiple-Input Multiple-Output (MIMO) relay communication systems, the main challenge is to get full knowledge of all channel matrices with minimal cost of signal handling at the relay node. In this paper, a low-complexity joint channel estimation scheme for two-way MIMO relay communication systems is proposed. Both users transmit orthogonal channel training signals to the relay node simultaneously. Then the relay amplifies the received signals by using designed amplification factors, and forwards the amplified signals to both users. The received signals at each user is formulated as a PARAllel FACtor (PARAFAC) model, and then the iterative and non-iterative fitting algorithms are derived to estimate the Channel State Information (CSI) knowledge of all links involved. Compared with existing schemes, the proposed scheme has the advantages of design flexibility and low complexity, and has higher estimation accuracy with a few number of channel training signals.
Two-way MIMO relay; Low-complexity; Channel State Information (CSI); Parallel factor
TN929.53
A
1009-5896(2017)12-2976-07
10.11999/JEIT170463
2017-05-16;
2017-09-16;
2017-11-02
通信作者:杜建和 dujianhe1@163.com
國(guó)家自然科學(xué)基金(61601414, 61561037),國(guó)家高技術(shù)研究發(fā)展計(jì)劃(2015AA01A705, 2014AA01A701)
: The National Natural Science Foundation of China (61601414, 61561037), The National High Technology Research and Development Program of China (2015AA01A705, 2014AA01A701)