搜索
第37卷第2期2003年2月西安交通大学学报JOURNALOFXIcANJIAOTONGUNIVERSITYVol.37l2Feb.2003徐超颖,杨国安,石永光,郑南宁(,710049,):提出了数字通信系统中在信道受到干扰时信道译码器检测或修正解调器送来错误信息的一种软件实现方案,该方案应用VisualC++610软件技术实现了卷积码编码器和维特比译码器功能,它不仅译码算法简单易实现,而且可以得到较大的编码增益,具有良好的纠错编码功能,是一种软件方法的前向纠错编码技术.实验结果表明:应用软判决维特比译码算法时的误码率低于应用硬判决算法的误码率,一般要比硬判决算法多大约2dB的增益;约束长度越大误码率越低,译码性能越好;在码率和约束长度不变时,硬判决算法的执行速度比软判决算法快.目前,该方案已应用于高精度网络彩色激光打印机中,并获得好评.:卷积码;维特比译码算法;硬判决;软判决:TN911:A:0253n987X(2003)02n0151n04SoftwareMethodofImplementConvolutionalCodeandItsViterbiDecodingAlgorithmXuChaoying,YangGuoan,ShiYongguang,ZhengNanning(SchoolofElectronicsandInformationEngineering,XicanJiaotongUniversity,Xican710049,China)Abstract:AsoftwareimplementationofachannelcodingtechnologyusingvisualC++610ispresented,whichralizesthefunctionsofconvolutionalcodingandViterbidecoding.Thissoftwarehastheadvantagesofasimpledecodingalgorithm,andalargercodinggainandprovidesabettererrorcorrectingperformance,whichbelongstoaforwarderrorcorrectioncodingtechnology.Theexper-imentalresultsshowthatthebiterrorrateofthesoft-decisionViterbidecoderislowerthanthatofhard-decisionwithalargergainof2dB.Thebiggertheconstraintlength,thelowerthebiterrorrateandthebetterthedecodingperformancebecomes.Whenthecoderateandconstraintlengthremainconstant,thespeedhard-decisionalgorithmishigherthanthatofsoft-decision.Atcurrentstage,theschemeproposedherehasbeenusedinthehigh-precisionnetworkcolorlaserprinterandwinsfavourablecommentsfromtheusers.Keywords:convotionalcode;Viterbidecodingalgorithm;hardwarejudging;softwarejudging,.2060,.,VLSI,.:,,,;,,[1,2].B(B),N(N\B).N-B,m.(K,B,N),K=m+1(ConstraintLength),m(MemoryLength);R=B/N(CodeRate).,,[3],.[4](Ad-ditiveWhiteGaussianNoise,AWGN),..:2002n04n29.:(1976~),,;(),,,,.:(60024301).12[5],:R=B/N=1/2,m=2,K=m+1=3,11121012,R=1/2K=3..0101110010100012.,0,0.1,0.,20,002.2,1.,0,,0,21002,1,2102,1.,112.3,0,1,0.,20102,1;2002,0,102.,,0011100001100111111000101100112.,(K,B,N),BK*N..,15,13,m=2.,m=2,0./0.,00111000011001111110001011001110112./0,m*B,.,0.,.,,(StateMachine).B=1,m=2,2m*B=4.21,20,0.10,0;11,102.1,,,.2,,.(3,1,2).2(K-1)*B=2m*B2B,2m*B,2B.,.表1下一时刻状态转换表=0=1000010010010100111110111表2输出状态表=0=10000110111001010011101102[6],.,6.(1)next_state[2m*B][2B],1,2m*B=2(K-1)*B,;2B,.,.(2)output[2m*B][2B],2,.(3)input[2m*B][2m*B],3,.(4)accum_err_metric[2m*B][2],4,2,.(5)state_history[2m*B][K*5+1],5,K*5+1,,.15237(6)state_sequence[K*5+1],6,.表3输入状态表002=0012=1102=2112=30020X1X012=10X1X102=2X0X1112=3X0X13.CPUPentiumÓ-1GHz,128MB,MicrosoftWin2000Pro,VisualC++610.1,BER,Eb/N0.,,2dB.K,2.,K,.RK,,7.10kB,Eb/N0210.(a)(3,1,2)(b)(5,1,3)(c)(3,2,3)1表4累积误差度量记录表t012345678910111213141516170020233334134332245201231223144142344210220213343141433211231211344342344表5状态历史记录表t01234567891011121314151617002000101101001010001012002233233223232220102000011101001101000112002232323223232200表6状态顺序记录表t012345678910111213141516170021233102121002101532,:(a)R=1/2(b)R=1/32K表7编译码时间的比较RK/s/s/s1/230.090.740.5250.162.341.5970.198.535.6990.2333.5522.021/330.140.940.6150.223.001.9170.2811.086.889013643123261522/330114217711665012241170241594,,.,BN,,.,,.,K,K,,.,.:[1]RheeMY.Error-correctingcodingtheory[M].NewYork:McGraw-HillPublishingCompany,1989.[2]MichelsonAM,LevesqueAH.Error-controltechniquesfordigitalcommunication[M].NewYork:JohnWiley&SonsInc,1985.[3].[M].:,1989.[4].[M].:,1989.[5].[M].:,1986.[6]FlemingC.AtutorialonconvolutionalcodingwithViterbidecoding,spectrumapplications[R/OL].http:Mpwl.net-com/~chip.f/Viterbi.html,2001n01n31.(苗凌)565西安交通大学学报6是一种自然科学类的学术理论性刊物.刊物水平的高低主要体现在其学术上的影响程度,而印刷编排规范只是其内容的表现形式.为了全面准确公正客观地评价和利用科技期刊,我国已制订了中国科技期刊评价指标体系,其中影响因子和被引频次是两个国际上通行的期刊评价指标.期刊影响因子和被引频次越大,它的学术影响力和作用也越大.影响因子是指该刊前两年发表论文在统计当年被引用的总次数除以该刊前两年发表论文总数.为了提高5西安交通大学学报6的学术水平,扩大学报的影响力,学报编辑部近年来做了大量的工作,采取了一系列行之有效的措施,使学报的各项评价指标有了较大提高,特别是影响因子和被引频次提高明显:2000年为01151和268次,2001年为01198和339次,2002年为01283和427次.(荆树蓉)15437