基于GA-ESN的移动音频抗丢包技术研究
|
摘要
随着移动和IP网络的发展,也带来了高品质移动音频的巨大需求,但由于网络拥塞,信道干扰和噪音等原因,实时音频通信不得不面对数据包丢失的问题,这严重影响了移动音频的质量。因此,能够提高丢包情况下移动音频质量的抗丢包技术变得日益重要。
回声状态网络(ESN)的预测能力可以用以提高移动音频质量。然而,ESN方法中存在输出反馈形成的闭合环路,可能导致ESN丧失渐进稳定性。本文提出了一种利用遗传算法(GA)改进回声状态网络(ESN)稳定性的方法,称为遗传算法回声状态网络(GA-ESN)。本方法采用相空间重构的方法对音频序列进行处理,再将音频送入GA-ESN进行预测,使GA-ESN获得更多的音频序列的信息,提高预测精度。这形成了一种结合混沌时间序列相空间重构方法的GA-ESN直接预测方法,在此基础上本文提出了基于GA-ESN的移动音频抗丢包算法。
本文进一步设计了一种基于GA-ESN的移动音频抗丢包算法框架,并在此框架下进行了音频信号的预测,即在25%的丢包环境下设计了一组实验,并用PEAQ和MOS的标准进行了客观和主观评价。实验结果表明,其效果优于工程中用的“错帧填零”方法。
With the development of mobile and IP networks, high-quality mobile audio is in great need. Due to the network congestion, channel interference and noise, real-time audio communications have to face the packet-loss issues which seriously affect the the quality of mobile audio. Therefore anti packet loss techniques become increasingly important to improve the quality of mobile audio in the packet-loss situations.
The echo state network (ESN) with the predictive capabilities can be used to improve the quality of mobile audio. However, the closed loop with feedback introduced in the ESN may lead to the loss of asymptotic stability. A genetic algorithm (GA) to improve the stability of ESN, called as genetic algorithm for echo state network (GA-ESN), is presented in the thesis. This algorithm reconstructs the audio sequence in the phase space based on the chaotic characteristics of the audio sequece, and predicts the sequence with the GA-ESN. As a result, an anti-packet-loss algorithm for mobile audio application based on the GA-ESN is designed and introduced.
A GA-ESN-based anti-packet loss algorithm for mobile audio framework is designed. Under this framework, some audio signal prediction experiments have been conducted under the 25% packet loss rate environment, the PEAQ and MOS are used to evaluate the results. The experiments show that the proposed algorithm is better than the method of "lost frame zero filling".
回声状态网络(ESN)的预测能力可以用以提高移动音频质量。然而,ESN方法中存在输出反馈形成的闭合环路,可能导致ESN丧失渐进稳定性。本文提出了一种利用遗传算法(GA)改进回声状态网络(ESN)稳定性的方法,称为遗传算法回声状态网络(GA-ESN)。本方法采用相空间重构的方法对音频序列进行处理,再将音频送入GA-ESN进行预测,使GA-ESN获得更多的音频序列的信息,提高预测精度。这形成了一种结合混沌时间序列相空间重构方法的GA-ESN直接预测方法,在此基础上本文提出了基于GA-ESN的移动音频抗丢包算法。
本文进一步设计了一种基于GA-ESN的移动音频抗丢包算法框架,并在此框架下进行了音频信号的预测,即在25%的丢包环境下设计了一组实验,并用PEAQ和MOS的标准进行了客观和主观评价。实验结果表明,其效果优于工程中用的“错帧填零”方法。
With the development of mobile and IP networks, high-quality mobile audio is in great need. Due to the network congestion, channel interference and noise, real-time audio communications have to face the packet-loss issues which seriously affect the the quality of mobile audio. Therefore anti packet loss techniques become increasingly important to improve the quality of mobile audio in the packet-loss situations.
The echo state network (ESN) with the predictive capabilities can be used to improve the quality of mobile audio. However, the closed loop with feedback introduced in the ESN may lead to the loss of asymptotic stability. A genetic algorithm (GA) to improve the stability of ESN, called as genetic algorithm for echo state network (GA-ESN), is presented in the thesis. This algorithm reconstructs the audio sequence in the phase space based on the chaotic characteristics of the audio sequece, and predicts the sequence with the GA-ESN. As a result, an anti-packet-loss algorithm for mobile audio application based on the GA-ESN is designed and introduced.
A GA-ESN-based anti-packet loss algorithm for mobile audio framework is designed. Under this framework, some audio signal prediction experiments have been conducted under the 25% packet loss rate environment, the PEAQ and MOS are used to evaluate the results. The experiments show that the proposed algorithm is better than the method of "lost frame zero filling".
引文
[1] Carle G,Biersack EW. Survey of Error Recovery Techniques for IP-Based Audio-Visual Multicast Applications. IEEE Network, 1997. 11(6): pp. 24-36
[2] Perkins C,Hodson O,Hardman V. A Survey of Packet-Loss Recovery Techniques for Streaming Audio. IEEE Network, Sept-Oct 1998. 12(5): pp. 40-48
[3] Ma HF,Xiaofeng H,Qian L, et al. Robust Audio and Speech Coding for Mobile and IP Network Applications. International conference on audio, language and image processing, 2008. Vol.1: pp. 261-266
[4]姜万通.混沌_分形与音乐关系的思索.沈阳音乐学院学报, 2004. 3: pp. 3-10
[5]姜万通.混沌·分形与音乐:音乐作品的混沌本质与分形研究初探. 2005,上海:上海音乐出版社
[6]姜万通.混沌分形与音乐. 2005,上海:上海音乐出版社
[7]王瑟.基于改进的回声状态神经网络的非线性预测. 2006,南京:南京工业大学
[8] Salihoglu U. Chaos in small recurrent neural networks:theoretical and practical studies, in Anne academic. 2003-2004, University Libre de Bruxelles: Brussels
[9] Jang JSR,Sun CT,Mizutani E. Neuro-Fuzzy and Soft Computing:a Computational Approach to Learning and Machine Intelligence. 1997, New York: Prentice Hall
[10] Jaeger H. Tutorial on training recurrent neural networks, covering BPPT, RTRL, EKF and the "echo state network" approach[GMD Report 159]. 2002, German National Research Center for Information Technology
[11] Jaeger H,Haass H. Harnessing nonlinearity: predicting chaotic systems and saving energy in wireless communication. Science, 2004. 304(5667): pp. 78-80
[12]雷英杰,张善文,李续武等. MATLAB遗传算法工具箱及应用. 2005,西安:西安电子科技大学出版
[13] Kennedy J,Eberhart R. Particle swarm optimization, in Proceedingsof IEEE International Conference on Neural Networks. 1995, IEEE: Perth, Australia. pp. 1942-1948
[14]吕振肃,侯志荣.自适应变异的粒子群优化算法.电子学报, 2004. 32(3): pp. 416-420
[15]王凌,刘波.微粒群优化与调度算法. 2008,北京:清华大学出版社
[16] Dorigo M,Caro GD. The ant colony optimization metaheuristic. in In New Ideas in Optimization. 1999. London,U.K: McGraw-Hill
[17] Dorigo M,Gambardella LM. Ant colony system: A cooperative learning approach to the traveling salesman problem. IEEE Transactions on Evolutionary Computation, 1997. 1(1): pp. 53-66
[18] Dorigo M,Maniezzo V,Colorni A. Ant system: Optimization by a colony of cooperating agents. IEEE Transactions on System,Man and Cybernetics,Part B, 1996. 26(1): pp. 29-41
[19]李士勇.蚁群算法及其应用. 2004,哈尔滨:哈尔滨工业大学出版社
[20]段海滨.蚁群算法原理及其应用. 2005,北京:科学出版社
[21] Maass W,Natschl?ger T,Markram H. Real-Time Computing Without Stable States: A New Framework for Neural Computation Based on Perturbations. Neural Computation, 2002. 14(11): pp. 2531-2560
[22]潘东毅,胡孝平,吴晓枫.基于遗传算法的H型钢檩条优化设计.湖州职业技术学院学报, 2009
[23]王抵修.地学数据分析中的相空间重构预测方法研究. 2008,长春:吉林大学
[24] Takens F. Determing strange attractors in turbulence. Lecture note in Math, 1981(898): pp. 361-381
[25]陈士华,陆君安.混沌动力学初步. 1998,武汉:武汉水利电力大学出版社
[26]张锁春.现代振荡反应的数学理论和数值方法. 1991,郑州:河南科学技术出版社
[27]张筑生.微分动力系统原理. 1987,北京:科学出版社
[28] Packard NH,Crutchfield JP,Farmer JD, et al. Geometry from a time series. Physical Review Letter, 1980. 45: pp. 712-716
[29] Grassberger P,Procaccia I. Measuring the strangeness of strange attractors. Physica D, 1983. 9: pp. 189-208
[30] Gibson JF,Farmer JD,Casdagli M, et al. An analytic approach to practical state space reconstruction. Physic D, 1992. 57: pp. 1-30
[31] Kugiumtzis D. State space reconstruction parameters in the analysis of chaotic time series the role of the time window length. Physica D, 1996. 95(1): pp. 13-28
[32] Kim HS,Eykholt RJ. Salas D.Nonlinear dynamics delay times,and embedding windows. Physica D, 1999. 127(1): pp. 48-60
[33] Otani M,Jones A. Automated embedding and creep phenomenon in chaotic time series[EB/OL]. 2000-10-14/2003-05-15 [cited; Available from: http://www.cs.cf.ac.uk/ec/aecp.html
[34]王东升,曹磊.混沌、分形及其应用. 1995,合肥:中国科技大学出版社
[35] Sauer T,Yorke JA. Casdagli M.Embedology. Journal of Statistical Physics, 1991. 65: pp. 579-616
[36] Ding MZ,Grebogi C,Ott E, et al. Estimating correlation dimension from chaotic time series:when does plateau onset occur. Physica D, 1993. 69: pp. 404-424
[37] Kennel MB,Brown R,Abarbanel HDI. Determining embedding dimension for phase-space reconstruction using geometrical construction. Physical Review A, 1992. 45(6): pp. 3403-3411
[38] Buzug T,Pfister G. Comparison of algorithms calculating optimal embedding parameters for delay time coordinates. Physica D, 1992. 58(1-4): pp. 127-137
[39] Aleksic Z. Estimating the embedding dimension. Physica D, 1991. 52(2-3): pp. 362-368
[40] Broomhead DS,King GP. Extracting qualitative dynamics from experimental data. Physica D, 1986. 20(2-3): pp. 217-236
[41] Vatutard R,Yiou P,Ghil M. Singular-spectrum analysis:atoolkit for short,noisy chaotic signals. Physica D, 1992. 58(1-4): pp. 95-126
[42] Mees AI,Rapp PE. Singular-value decomposition and embedding dimension. Physical Review A, 1987. 36(1): pp. 340-346
[43] Fraster AM. Information and entropy in strange attractors. IEEE Transactions on IT, 1989. IT-35(2): pp. 245-262
[44] Hong P,Peterson.C. Finding the embedding dimension and variable dependencies in time series. Neural Computation 1994. 6(3): pp. 509-518
[45]龚云帆,徐健学.混沌信号和噪声.信号处理, 1997. 13(2): pp. 112-147
[46] Palus M,Dvorak I. Singular-value decomposition in attractorreconstruction:pitfalls and precaution. Physica D, 1992. 55(1-2): pp. 221-234
[47]林嘉宇,黄芝平,王跃科等.语音信号相空间重构中嵌入维数的选择.电子科学学刊, 1999. 21(6): pp. 735-742
[48]马红光,李夕海,王国华等.相空间重构中嵌入维和时间延迟的选择.西安交通大学学报, 2004. 38(4): pp. 335-338
[49]林嘉宇,王跃科,黄芝平等.语音信号相空间重构中的时间延迟的选择—复自相关法.信号处理, 1999. 15(3): pp. 220-225
[50] Rosenstein MT,Collins JJ,Deluca CJ. Reconstruction expansion as a geometry-based framework for choosing proper delay times. Physica D, 1994. 73: pp. 82-98
[51] Fraser AM. Swinney H I.Independent coordinates for strange attractors from mutual information. Physical Review A, 1986. 33: pp. 1134-1140
[52]韩敏,史志伟,郭伟.储备池状态空间重构与混沌时间序列预测.物理学报, 2005. 56(1): pp. 44-46
[53]章鑫.抗丢包宽带音频编码算法的研究和实现. 2008,西安:西安电子科技大学
[54] ITU-T Recommendation P.862.Perceptual evaluation of speech quality (PESQ),an objective method for end-to-end speech quality assessment of narrowband telephone networks and speech codecs. 2001
[55] ITU-R Recommendation BS.1387.Method for objective measurements of perceived audio quality(PEAQ). 1998-2001
[2] Perkins C,Hodson O,Hardman V. A Survey of Packet-Loss Recovery Techniques for Streaming Audio. IEEE Network, Sept-Oct 1998. 12(5): pp. 40-48
[3] Ma HF,Xiaofeng H,Qian L, et al. Robust Audio and Speech Coding for Mobile and IP Network Applications. International conference on audio, language and image processing, 2008. Vol.1: pp. 261-266
[4]姜万通.混沌_分形与音乐关系的思索.沈阳音乐学院学报, 2004. 3: pp. 3-10
[5]姜万通.混沌·分形与音乐:音乐作品的混沌本质与分形研究初探. 2005,上海:上海音乐出版社
[6]姜万通.混沌分形与音乐. 2005,上海:上海音乐出版社
[7]王瑟.基于改进的回声状态神经网络的非线性预测. 2006,南京:南京工业大学
[8] Salihoglu U. Chaos in small recurrent neural networks:theoretical and practical studies, in Anne academic. 2003-2004, University Libre de Bruxelles: Brussels
[9] Jang JSR,Sun CT,Mizutani E. Neuro-Fuzzy and Soft Computing:a Computational Approach to Learning and Machine Intelligence. 1997, New York: Prentice Hall
[10] Jaeger H. Tutorial on training recurrent neural networks, covering BPPT, RTRL, EKF and the "echo state network" approach[GMD Report 159]. 2002, German National Research Center for Information Technology
[11] Jaeger H,Haass H. Harnessing nonlinearity: predicting chaotic systems and saving energy in wireless communication. Science, 2004. 304(5667): pp. 78-80
[12]雷英杰,张善文,李续武等. MATLAB遗传算法工具箱及应用. 2005,西安:西安电子科技大学出版
[13] Kennedy J,Eberhart R. Particle swarm optimization, in Proceedingsof IEEE International Conference on Neural Networks. 1995, IEEE: Perth, Australia. pp. 1942-1948
[14]吕振肃,侯志荣.自适应变异的粒子群优化算法.电子学报, 2004. 32(3): pp. 416-420
[15]王凌,刘波.微粒群优化与调度算法. 2008,北京:清华大学出版社
[16] Dorigo M,Caro GD. The ant colony optimization metaheuristic. in In New Ideas in Optimization. 1999. London,U.K: McGraw-Hill
[17] Dorigo M,Gambardella LM. Ant colony system: A cooperative learning approach to the traveling salesman problem. IEEE Transactions on Evolutionary Computation, 1997. 1(1): pp. 53-66
[18] Dorigo M,Maniezzo V,Colorni A. Ant system: Optimization by a colony of cooperating agents. IEEE Transactions on System,Man and Cybernetics,Part B, 1996. 26(1): pp. 29-41
[19]李士勇.蚁群算法及其应用. 2004,哈尔滨:哈尔滨工业大学出版社
[20]段海滨.蚁群算法原理及其应用. 2005,北京:科学出版社
[21] Maass W,Natschl?ger T,Markram H. Real-Time Computing Without Stable States: A New Framework for Neural Computation Based on Perturbations. Neural Computation, 2002. 14(11): pp. 2531-2560
[22]潘东毅,胡孝平,吴晓枫.基于遗传算法的H型钢檩条优化设计.湖州职业技术学院学报, 2009
[23]王抵修.地学数据分析中的相空间重构预测方法研究. 2008,长春:吉林大学
[24] Takens F. Determing strange attractors in turbulence. Lecture note in Math, 1981(898): pp. 361-381
[25]陈士华,陆君安.混沌动力学初步. 1998,武汉:武汉水利电力大学出版社
[26]张锁春.现代振荡反应的数学理论和数值方法. 1991,郑州:河南科学技术出版社
[27]张筑生.微分动力系统原理. 1987,北京:科学出版社
[28] Packard NH,Crutchfield JP,Farmer JD, et al. Geometry from a time series. Physical Review Letter, 1980. 45: pp. 712-716
[29] Grassberger P,Procaccia I. Measuring the strangeness of strange attractors. Physica D, 1983. 9: pp. 189-208
[30] Gibson JF,Farmer JD,Casdagli M, et al. An analytic approach to practical state space reconstruction. Physic D, 1992. 57: pp. 1-30
[31] Kugiumtzis D. State space reconstruction parameters in the analysis of chaotic time series the role of the time window length. Physica D, 1996. 95(1): pp. 13-28
[32] Kim HS,Eykholt RJ. Salas D.Nonlinear dynamics delay times,and embedding windows. Physica D, 1999. 127(1): pp. 48-60
[33] Otani M,Jones A. Automated embedding and creep phenomenon in chaotic time series[EB/OL]. 2000-10-14/2003-05-15 [cited; Available from: http://www.cs.cf.ac.uk/ec/aecp.html
[34]王东升,曹磊.混沌、分形及其应用. 1995,合肥:中国科技大学出版社
[35] Sauer T,Yorke JA. Casdagli M.Embedology. Journal of Statistical Physics, 1991. 65: pp. 579-616
[36] Ding MZ,Grebogi C,Ott E, et al. Estimating correlation dimension from chaotic time series:when does plateau onset occur. Physica D, 1993. 69: pp. 404-424
[37] Kennel MB,Brown R,Abarbanel HDI. Determining embedding dimension for phase-space reconstruction using geometrical construction. Physical Review A, 1992. 45(6): pp. 3403-3411
[38] Buzug T,Pfister G. Comparison of algorithms calculating optimal embedding parameters for delay time coordinates. Physica D, 1992. 58(1-4): pp. 127-137
[39] Aleksic Z. Estimating the embedding dimension. Physica D, 1991. 52(2-3): pp. 362-368
[40] Broomhead DS,King GP. Extracting qualitative dynamics from experimental data. Physica D, 1986. 20(2-3): pp. 217-236
[41] Vatutard R,Yiou P,Ghil M. Singular-spectrum analysis:atoolkit for short,noisy chaotic signals. Physica D, 1992. 58(1-4): pp. 95-126
[42] Mees AI,Rapp PE. Singular-value decomposition and embedding dimension. Physical Review A, 1987. 36(1): pp. 340-346
[43] Fraster AM. Information and entropy in strange attractors. IEEE Transactions on IT, 1989. IT-35(2): pp. 245-262
[44] Hong P,Peterson.C. Finding the embedding dimension and variable dependencies in time series. Neural Computation 1994. 6(3): pp. 509-518
[45]龚云帆,徐健学.混沌信号和噪声.信号处理, 1997. 13(2): pp. 112-147
[46] Palus M,Dvorak I. Singular-value decomposition in attractorreconstruction:pitfalls and precaution. Physica D, 1992. 55(1-2): pp. 221-234
[47]林嘉宇,黄芝平,王跃科等.语音信号相空间重构中嵌入维数的选择.电子科学学刊, 1999. 21(6): pp. 735-742
[48]马红光,李夕海,王国华等.相空间重构中嵌入维和时间延迟的选择.西安交通大学学报, 2004. 38(4): pp. 335-338
[49]林嘉宇,王跃科,黄芝平等.语音信号相空间重构中的时间延迟的选择—复自相关法.信号处理, 1999. 15(3): pp. 220-225
[50] Rosenstein MT,Collins JJ,Deluca CJ. Reconstruction expansion as a geometry-based framework for choosing proper delay times. Physica D, 1994. 73: pp. 82-98
[51] Fraser AM. Swinney H I.Independent coordinates for strange attractors from mutual information. Physical Review A, 1986. 33: pp. 1134-1140
[52]韩敏,史志伟,郭伟.储备池状态空间重构与混沌时间序列预测.物理学报, 2005. 56(1): pp. 44-46
[53]章鑫.抗丢包宽带音频编码算法的研究和实现. 2008,西安:西安电子科技大学
[54] ITU-T Recommendation P.862.Perceptual evaluation of speech quality (PESQ),an objective method for end-to-end speech quality assessment of narrowband telephone networks and speech codecs. 2001
[55] ITU-R Recommendation BS.1387.Method for objective measurements of perceived audio quality(PEAQ). 1998-2001