靳 一 吳樂(lè)南 馮 熳 鄧 蕾
(東南大學(xué)信息科學(xué)與工程學(xué)院,南京 210096)
隨機(jī)極性MCP-EBPSK傳輸性能
靳 一 吳樂(lè)南 馮 熳 鄧 蕾
(東南大學(xué)信息科學(xué)與工程學(xué)院,南京 210096)
為提高隨機(jī)極性的修正連續(xù)相位擴(kuò)展二元相移鍵控(MCP-EBPSK)調(diào)制信號(hào)的傳輸性能,對(duì)比了AWGN信道上自適應(yīng)門(mén)限判決、BP神經(jīng)網(wǎng)絡(luò)和支持向量機(jī)(SVM)判決3種解調(diào)方式的解調(diào)性能,并引入了RS碼和規(guī)則LDPC碼來(lái)進(jìn)行改善.根據(jù)隨機(jī)極性MCP-EBPSK調(diào)制原理和數(shù)字沖擊濾波器特殊濾波機(jī)理,設(shè)計(jì)了基于自適應(yīng)門(mén)限判決、BP神經(jīng)網(wǎng)絡(luò)和SVM的解調(diào)器.同時(shí),在這3種解調(diào)方式性能均不佳時(shí),對(duì)RS碼和規(guī)則LDPC編碼的隨機(jī)極性MCP-EBPSK傳輸性能進(jìn)行了仿真.仿真結(jié)果表明:在AWGN信道上,自適應(yīng)門(mén)限判決、BP神經(jīng)網(wǎng)絡(luò)和SVM三種解調(diào)方式的解調(diào)性能相差不大;當(dāng)采用自適應(yīng)門(mén)限判決且誤碼率為10-4時(shí),RS碼和規(guī)則LDPC碼的編碼增益分別約為4 dB和7 dB.因此,引入信道編碼可顯著改善隨機(jī)極性MCP-EBPSK的傳輸性能.
修正連續(xù)相位擴(kuò)展二元相移鍵控;自適應(yīng)門(mén)限判決;BP神經(jīng)網(wǎng)絡(luò);支持向量機(jī);數(shù)字沖擊濾波器;編碼增益
隨著信息社會(huì)的不斷發(fā)展,人們對(duì)信息傳輸速率和可靠性的要求越來(lái)越高,直接導(dǎo)致空中的無(wú)線(xiàn)電頻譜資源愈加緊張.如何高效合理地利用有限的頻譜資源已成為熱門(mén)研究課題.從以前的高階調(diào)制到目前的多天線(xiàn)技術(shù)(MIMO)[1]、認(rèn)知無(wú)線(xiàn)電技術(shù)(CR)[2]均在一定程度上提高了頻譜效率,卻并未從根本上解決頻譜資源不足的問(wèn)題.超窄帶(UNB)調(diào)制通過(guò)對(duì)通信體制的變革從根本上來(lái)解決頻譜效率的低下問(wèn)題.文獻(xiàn)[3-8]都對(duì)此技術(shù)進(jìn)行了深入的研究.然而,超窄帶能否真正實(shí)用則取決于其傳輸性能的優(yōu)劣.
本文對(duì)隨機(jī)極性MCP-EBPSK調(diào)制在加性白高斯噪聲(AWGN)信道的解調(diào)性能進(jìn)行研究,并分別引入RS碼[9]和規(guī)則 LDPC 碼[10-12]來(lái)改善其傳輸性能.首先,介紹隨機(jī)極性MCP-EBPSK調(diào)制的基本原理.其次,詳細(xì)闡述數(shù)字沖擊濾波器[13]的特殊濾波機(jī)理,并給出基于幅度差異和波形特征差異的兩大類(lèi)解調(diào)器結(jié)構(gòu).然后,給出信道編碼的隨機(jī)極性MCP-EBPSK通信系統(tǒng)框圖.最后,對(duì)比和分析基于幅度差異的自適應(yīng)門(mén)限判決和基于波形特征的BP神經(jīng)網(wǎng)絡(luò)、支持向量機(jī)(SVM)的解調(diào)性能,并在AWGN信道分別仿真RS和LDPC編碼的隨機(jī)極性MCP-EBPSK系統(tǒng)性能.
隨機(jī)極性MCP-EBPSK調(diào)制[5]通過(guò)在隨機(jī)極性CP-EBPSK調(diào)制中引入一個(gè)功率譜形狀的調(diào)節(jié)系數(shù)η<1,來(lái)改善功率譜結(jié)構(gòu),使調(diào)制信號(hào)的能量更加向載頻和功率譜主瓣集中,有利于提升能量利用率,其表達(dá)式為
式中,ωc為載波角頻率;T=2π/ωc為載波周期;碼元“0”持續(xù)N個(gè)載波周期,碼元“1”的相位調(diào)制持續(xù)K個(gè)載波周期,K和N均為整數(shù)以保證整周期調(diào)制;ξ∈{-1,1}代表了相位調(diào)制的極性;0<Δ<1為調(diào)相指數(shù);η為功率譜形狀調(diào)節(jié)系數(shù),由N,K,Δ和η共同構(gòu)成了改變信號(hào)帶寬、傳輸效率和解調(diào)性能的一組“調(diào)制指數(shù)”.對(duì)于“0”碼元,發(fā)送的調(diào)制波形為f0(t),“1”碼元?jiǎng)t發(fā)送由“ξΔ”共同決定的調(diào)制波形f1(t),由此可在嚴(yán)格的-60 dB帶寬下得到很高的頻譜利用率.
數(shù)字沖擊濾波器[13]是一種具有“陷波-選頻”特性的IIR濾波器,在極窄的通帶內(nèi)可將隨機(jī)極性MCP-EBPSK調(diào)制信號(hào)的微弱相位變化轉(zhuǎn)化為幅度沖擊,且沖擊包絡(luò)不受相位極性影響,增強(qiáng)了調(diào)制波形的差異,有利于門(mén)限判決和簡(jiǎn)化接收機(jī)結(jié)構(gòu).在此,選用1對(duì)共軛零點(diǎn)和3對(duì)共軛極點(diǎn)的數(shù)字沖擊濾波器,其傳輸函數(shù)為
式中
當(dāng)載頻 fc=21.4 MHz,采樣頻率 fs=214 MHz,N=10,K=2,調(diào)相指數(shù) Δ =0.1,η =1/2時(shí),隨機(jī)極性的MCP-EBPSK調(diào)制信號(hào)經(jīng)過(guò)上述數(shù)字沖擊濾波器,得到了如圖1所示的沖擊波形和輸出包絡(luò).從圖1可看出,數(shù)字沖擊濾波器的輸出包絡(luò)不受“1”碼元的調(diào)制相位極性的影響,可在一定程度上放大“0”和“1”調(diào)制波形的差異,有利于直接門(mén)限判決或采用基于分類(lèi)器思想的BP神經(jīng)網(wǎng)絡(luò)、SVM 解調(diào).
圖1 隨機(jī)極性MCP-EBPSK沖擊波形和包絡(luò)
依據(jù)2.1節(jié)所述沖擊濾波機(jī)理,可設(shè)計(jì)2類(lèi)不同的解調(diào)器:①根據(jù)沖擊包絡(luò)幅度的差異,設(shè)計(jì)基于自適應(yīng)門(mén)限判決的解調(diào)器;② 依據(jù)沖擊包絡(luò)波形特征的差異,設(shè)計(jì)基于BP神經(jīng)網(wǎng)絡(luò)、SVM等分類(lèi)器的解調(diào)器.2種解調(diào)器結(jié)構(gòu)分別如圖2和圖3所示.其中,圖2和圖3中的預(yù)處理均為取絕對(duì)值和低通濾波,即通常的包絡(luò)檢波.
圖2 基于自適應(yīng)門(mén)限判決的解調(diào)器
圖3 基于BP神經(jīng)網(wǎng)絡(luò)/SVM的解調(diào)器
圖4為信道編碼的隨機(jī)極性MCP-EBPSK通信系統(tǒng)框圖.本節(jié)首先在AWGN信道對(duì)比圖2和圖3所示2種解調(diào)器的解調(diào)性能,然后引入RS碼和規(guī)則LDPC碼來(lái)改善傳輸性能.仿真參數(shù)均與2.1節(jié)保持一致.
圖4 隨機(jī)極性MCP-EBPSK通信系統(tǒng)
在訓(xùn)練信噪比為29 dB,測(cè)試信噪比為27~31 dB時(shí),分別在訓(xùn)練樣本數(shù)為1 000,2 000和3 000的情況下,對(duì)基于BP神經(jīng)網(wǎng)絡(luò)的解調(diào)器進(jìn)行了仿真,結(jié)果如圖5(a)所示.同時(shí),在訓(xùn)練樣本數(shù)為2 000時(shí),分別在訓(xùn)練信噪比為27,29和31 dB的情況下,得到了如圖5(b)所示的仿真結(jié)果.然后,在訓(xùn)練樣本數(shù)為2 000和訓(xùn)練信噪比為29 dB時(shí),分別在特征點(diǎn)數(shù)為30,40,50和100的情況下進(jìn)行了仿真,得到了如圖5(c)所示的仿真結(jié)果.從圖5(a)~(c)可以看出,訓(xùn)練樣本數(shù)和特征點(diǎn)數(shù)并非越多越好,當(dāng)訓(xùn)練樣本數(shù)為2 000,特征點(diǎn)數(shù)為40,而訓(xùn)練信噪比取測(cè)試范圍的中間值29 dB時(shí),BP神經(jīng)網(wǎng)絡(luò)的解調(diào)性能較佳.圖5(d)給出了該組訓(xùn)練參數(shù)下BP神經(jīng)網(wǎng)絡(luò)和自適應(yīng)門(mén)限判決的解調(diào)性能對(duì)比,可看出僅在信噪比超過(guò)28 dB時(shí),BP神經(jīng)網(wǎng)絡(luò)的解調(diào)性能才優(yōu)于自適應(yīng)門(mén)限判決,且信噪比的最大提升量約為0.5 dB.
在與圖5(a)~(c)相同的訓(xùn)練條件下對(duì)基于SVM的解調(diào)器進(jìn)行仿真,結(jié)果如圖6(a)~(c)所示.可看出,與BP神經(jīng)網(wǎng)絡(luò)類(lèi)似,訓(xùn)練樣本數(shù)和特征點(diǎn)數(shù)并非越多越好,當(dāng)訓(xùn)練樣本數(shù)為2 000,特征點(diǎn)數(shù)為40,而訓(xùn)練信噪比取測(cè)試范圍的中間值29 dB時(shí),SVM的解調(diào)性能較佳.圖6(d)給出了該組訓(xùn)練參數(shù)下SVM和自適應(yīng)門(mén)限判決的解調(diào)性能對(duì)比,可看出僅在信噪比超過(guò)28.5 dB時(shí),SVM 的解調(diào)性能才優(yōu)于自適應(yīng)門(mén)限判決,且信噪比的最大提升量約為0.5 dB.
由3.1節(jié)和3.2節(jié)可知,以上3種解調(diào)方式的解調(diào)性能均不佳,這是由于隨機(jī)極性MCP-EBPSK極小的相位跳變所產(chǎn)生的沖擊包絡(luò)不太明顯.本節(jié)分別引入不同碼長(zhǎng)的RS碼和規(guī)則LDPC碼對(duì)整個(gè)通信系統(tǒng)進(jìn)行仿真,仿真碼元數(shù)為1.0×107個(gè),得到了如圖7和圖8所示的誤碼率曲線(xiàn)(未采用交織和解交織,因其在AWGN信道中作用有限).圖7給出了 RS(15,9),RS(31,19),RS(31,23),RS(63,51)和RS(255,223)等5種不同碼長(zhǎng)的 RS碼和無(wú)信道編碼時(shí)的誤碼率曲線(xiàn),可看出,RS(31,19)碼性能最好.與非編碼系統(tǒng)相比,當(dāng)誤碼率為10-4時(shí),RS(31,19)碼可獲得約4 dB 的編碼增益.同時(shí),采用碼率為1/2和碼長(zhǎng)為2 000的規(guī)則LDPC碼作為信道編碼,并借鑒文獻(xiàn)[14]中利用SVM的后驗(yàn)概率模型估計(jì)EBPSK沖擊濾波輸出包絡(luò)的后驗(yàn)概率的思想,利用和積算法譯碼迭代50次,得到了如圖8所示的誤碼率曲線(xiàn).可看出,當(dāng)誤碼率為10-4時(shí),約有7 dB的編碼增益.
圖5 基于BP神經(jīng)網(wǎng)絡(luò)的解調(diào)器仿真結(jié)果
圖6 基于SVM的解調(diào)器仿真結(jié)果
圖7 不同碼長(zhǎng)RS碼的解調(diào)性能
圖8 1/2碼率規(guī)則LDPC碼的解調(diào)性能
對(duì)于AWGN信道,現(xiàn)有的隨機(jī)極性MCP-EBPSK解調(diào)器,無(wú)論是基于幅度差異的自適應(yīng)門(mén)限判決,還是基于波形特征的BP神經(jīng)網(wǎng)絡(luò)、SVM,其解調(diào)性能均不佳.在隨機(jī)極性MCP-EBPSK通信系統(tǒng)中引入RS碼和LDPC碼,可獲得較大的編碼增益.當(dāng)BER 為10-4時(shí),RS(31,19)碼可獲得約4 dB的編碼增益;碼率為1/2、碼長(zhǎng)為2 000的規(guī)則LDPC碼可獲得約7 dB的編碼增益,從而將解調(diào)所需的信噪比從超過(guò)30 dB降低到25 dB以下,有望用于具有較高信噪比的光纖通信或數(shù)字電視信號(hào)的電纜傳輸.
[1]Alamouti S M.A simple transmit diversity technique for wireless communications[J].IEEE Journal on Selected Areas in Communications,1998,16(8):1451-1458.
[2] Mitola J,Maguire G Jr.Cognitive radio:making software radios more personal[J].IEEE Personal Communication Magazine,1999,6(4):13-18.
[3] Walker H R.Ultra narrow band modulation textbook[M/OL].(2009)[2011-11-20].http://www.vmsk.org/.
[4] Wu Lenan,F(xiàn)eng Man,Qi Chenhao,et al.Recent patents on ultra narrow band modulations[J].Recent Patents on Signal Processing,2011,1(1):36-47.
[5]靳一,吳樂(lè)南,馮熳,等.一種隨機(jī)極性MCP-EBPSK調(diào)制解調(diào)器[J].電子與信息學(xué)報(bào),2012,34(7):1647-1652.
Jin Yi,Wu Lenan,F(xiàn)eng Man,et al.An MCP-EBPSK modem with random-polar[J].Journal of Electronics&Information Technology,2012,34(7):1647-1652.(in Chinese)
[6] Zheng Guoxin,F(xiàn)eng Jinzhen,Jia Minghua.Very minimum chirp keying as a novel ultra narrow band communication scheme[C]//The6th International Conference on Information,Communication&Signal Processing.Singapore,2007:1-3.
[7] Zheng Guoxin,Yang Weiying.The orthogonal very minimum chirp keying(OVMCK)modulations with very high bandwidth efficiency[C]//Proc of2008IEEE International Symposium of Antennas and Propagation Society.San Diego,USA,2008:3182-3185.
[8] Li Bin,Zhou Zheng,Zou Weixia.RPPK modulation with high data rates[J].Science China:Information Sciences,2010,53(2):344-354.
[9]Yar K P,Yoo D S,Stark W.Performance of RS coded M-ary modulation with and without symbol overlapping[J].IEEE Transactions on Communications,2008,56(3):445-453.
[10]Gallager R G.Low-density parity-check codes[J].IRE Transactions on Information Theory,1962,8(1):21-28.
[11]Abu-Surra S,Divsalar D,Ryan W E.Enumerators for protograph-based ensembles of LDPC and generalized LDPC codes[J].IEEE Transactions on Information Theory,2011,57(2):858-886.
[12] Rybin P,Zyablov V.Asymptotic estimation of error fraction corrected by binary LDPC code[C]//2011IEEE International Symposium on Information Theory Proceedings.St.Petersburg,NJ,USA,2011:351-355.
[13]馮熳,高鵬,吳樂(lè)南.超窄帶調(diào)制信號(hào)的特殊濾波分析與仿真[J].東南大學(xué)學(xué)報(bào):自然科學(xué)版,2010,40(2):227-230.
Feng Man,Gao Peng,Wu Lenan.Analysis and simulation of special filtering based on ultra narrow band modulated signal[J].Journal of Southeast University:Natural Science Edition,2010,40(2):227-230.(in Chinese)
[14]陳賢卿,吳樂(lè)南.EBPSK解調(diào)器中利用SVM概率輸出的LDPC譯碼[J].信號(hào)處理,2011,27(9):1286-1290.
Chen Xianqing,Wu Lenan.LDPC decoding via SVM probability output in EBPSK demodulator[J].Signal Processing,2011,27(9):1286-1290.(in Chinese)
Transmission performance of MCP-EBPSK with random polar
Jin Yi Wu Lenan Feng Man Deng Lei
(School of Information Science and Engineering,Southeast University,Nanjing 210096,China)
In order to improve the transmission performance of random polar and modified extended binary phase shift keying with continuous phase(MCP-EBPSK)modulated signal,the demodulation performance of the adaptive threshold discrimination,BP neural network,and support vector machine(SVM)in additive white Gaussian noise(AWGN)channel is compared,then Reed-Solomon(RS)code and regular low-density parity-check(LDPC)code are introduced.According to the principle of MCP-EBPSK with random polar and the special filtering mechanism of digital impacting filter,the demodulators are designed based on adaptive threshold discrimination,BP neural network,and SVM.Meanwhile,when the demodulation performance of these demodulators is poor,the RS and regular LDPC coded MCP-EBPSK with random polar is also simulated.Simulation results show that in AWGN channel,the demodulation performance of adaptive threshold discrimination,BP neural network and SVM is almost the same;when adaptive threshold discrimination is used and BER is 10-4,the coding gain of RS code and regular LDPC is about 4 dB and 7 dB,respectively.Therefore,channel code can be introduced to significantly improve the transmission performance of MCPEBPSK with random polar.
modified extended binary phase shift keying with continuous phase(MCP-EBPSK);adaptive threshold discrimination;BP neural network;support vector machine;digital impacting filter;coding gain
TN911.25
A
1001-0505(2012)06-1031-05
10.3969/j.issn.1001 -0505.2012.06.002
2012-05-08.
靳一(1984—),男,博士生;吳樂(lè)南(聯(lián)系人),男,博士,教授,博士生導(dǎo)師,wuln@seu.edu.cn.
國(guó)家自然科學(xué)基金資助項(xiàng)目(60872075).
靳一,吳樂(lè)南,馮熳,等.隨機(jī)極性 MCP-EBPSK傳輸性能[J].東南大學(xué)學(xué)報(bào):自然科學(xué)版,2012,42(6):1031-1035.[doi:10.3969/j.issn.1001 -0505.2012.06.002]