基于DM642的多媒体通信系统中语音通信子系统的研究与实现
摘要
本文主要研究基于嵌入式平台多媒体通信中的语音通信子系统。重点分析了语音通信系统的工作原理和质量标准,针对课题需要和服务质量的要求,提出了基于嵌入式平台的语音通信系统。并在DM642硬件平台上设计并实现了多点之间的语音通信系统。通过测试,验证了本系统能满足课题的需求。
文中实现了基于嵌入式平台的音频通信系统的构建、音频实时通信技术的研究、基于SIP协议栈的多方语音终端之间控制信令的设计与实现、音频驱动的设计与实现、基于RF5的语音编解码算法的CELL封装、网络模块的设计与实现以及多任务通信等关键技术。根据系统存在的问题提出了相关的性能保证机制,并给出实现策略。经过验证,改进后的系统有很强的稳定性和易扩展性。最后,对全文的工作进行了总结,并指出了今后有待进一步完善的地方。
With the development of network communications and multimedia technologies, the application of delivering audio and video over Internet is getting more and more popular, and today come forth various audio and video systems. Moreover, with the rapid improvement of DSPs processing ability, audio and video processing platform quickly trends to miniaturization. The research on audio and video real-time communications based on embedded DSPs system becomes a hotspot.
This paper mainly research on audio communication system which is part of multimedia communication. Firstly,To master the status in quo about audio communication technologies through consult information, study and analysis existented problem at present about the audio communication technologies and related solutions. Secondly, build the audio communication system based on the DM642 platform, the keystone and difficulty including these aspects below, building of the system based on the hardware platform, replant and optimize of the G.729A coder and decoder algorithm, the design and realization of audio communication control signaling based on SIP stack, the transport technologies on the Internet of the speech signal, the design of multi-tasks and task synchronize mechanism and so on. Try to master the core technology of audio communication via achieve the system, and to get ready for explore audio and video systems in the future.
This system is built on TI DM642EVM platform. TI DM642 is a kind of DSP chip designed for multimedia applications. Its core clock is up to 600MHz, and it has 8 parallel operation units, its processing ability is up to 4800MIPS. For multimedia applications, DM642EVM integrates 3 configurable video ports, McASP for audio applications, 10/100Mb/s Ethernet MAC and so on.
On the realization of the system, this system use CCS, adopt DSP/BIOS real-time OS, BSL (Board supported library), DDK (Driver develop kit), RF5 (Reference framework 5), CSL (Chip supported library) and other supported module to develop the software. Using of these module improving the programme efficiency, shorten the system implement period. And the system using C language to design and realize the audio communication system.
This system realized the point to point and point to multi-point audio communication system, each endpoint of the system is based on the DM642 platform. Each endpoint is consisted of two parts, signaling module and media module. Signaling module includes three tasks, signaling send and receives task and uart task. Media module including six tasks, audio capture task, audio coder and decoder task, audio send and receive task, and audio player task. Tasks of signaling module are synchronized by extern variable; Tasks of media module are synchronized using SCOM mechanism.
SCOM mechanism is a kind of synchonize mechanism which is offered by RF5. It build absolute message unilateralism channel for every task which need to synchronize with other tasks to realize the synchronize control, and use the message structure to realize the share and mutex of the resource. The disposal of the audio data between tasks in this system uses SCOM to realize synchronization. This system is to develop the software on the hardware platform, to accomplish the design and realization of nine modules talked above. The system is devided into two parts, signaling module and media module.
The signaling module is consisting of keyboard and UART. In this system, using the keyboard to simulate dial to trigger signaling, to achieve call setup, media negotiation and call realease and so on.
The media module is consisting of audio codec, DM642 and ethernet chips. After the setup of the signaling, the media modules achieve audio capture, process, and transport and so on. The principle of the system is this, firstly,control the AIC23 audio coder and decoder chip and audio port to realize the audio capture, keep the audio data captured in the capture buffer and compress the captured audio data using G.729A algorithm. Secondly, use the SCOM to transport audio data between tasks and use ethernet to complete the audio communication between DM642 and PC or DM642 and DM642. Finally, the receive endpoint play the audio after a series of reverse operation to realize the IP phone system based on the embed system.
At last, this pater tests the performance of the system and proposes related methods to improve the Qos performance of the system. First, the buffer mechanism of send and receive tasks, Second, audio coder and decoder algorithm. Then study these two methods separately and give the test result. This system adopts G.729A audio coder and decoder algorithm which is popular today. The algorithm can accomplish the speech compression in very low bit rate, suit for the demand of the IP phone today. In this system we wrapper the algorithm in a CELL according to the RF5 and optimize it according to the characters of the hardware platform. This paper also compare the former and after performance and prove that the performance after optimize is better than before.
According to what we have talked above, we can concluded that this article researches some key technologies about the audio communication based on embed system, designs and achieves a audio communication system and point to point and point to multi-point control signaling based on DM642 hardware platform.However realized the improvement on the problems exist in the system and proved that the effect is very well. For the sake of the time, this system is not perfect, and the author will improve it during afterward works.
文中实现了基于嵌入式平台的音频通信系统的构建、音频实时通信技术的研究、基于SIP协议栈的多方语音终端之间控制信令的设计与实现、音频驱动的设计与实现、基于RF5的语音编解码算法的CELL封装、网络模块的设计与实现以及多任务通信等关键技术。根据系统存在的问题提出了相关的性能保证机制,并给出实现策略。经过验证,改进后的系统有很强的稳定性和易扩展性。最后,对全文的工作进行了总结,并指出了今后有待进一步完善的地方。
With the development of network communications and multimedia technologies, the application of delivering audio and video over Internet is getting more and more popular, and today come forth various audio and video systems. Moreover, with the rapid improvement of DSPs processing ability, audio and video processing platform quickly trends to miniaturization. The research on audio and video real-time communications based on embedded DSPs system becomes a hotspot.
This paper mainly research on audio communication system which is part of multimedia communication. Firstly,To master the status in quo about audio communication technologies through consult information, study and analysis existented problem at present about the audio communication technologies and related solutions. Secondly, build the audio communication system based on the DM642 platform, the keystone and difficulty including these aspects below, building of the system based on the hardware platform, replant and optimize of the G.729A coder and decoder algorithm, the design and realization of audio communication control signaling based on SIP stack, the transport technologies on the Internet of the speech signal, the design of multi-tasks and task synchronize mechanism and so on. Try to master the core technology of audio communication via achieve the system, and to get ready for explore audio and video systems in the future.
This system is built on TI DM642EVM platform. TI DM642 is a kind of DSP chip designed for multimedia applications. Its core clock is up to 600MHz, and it has 8 parallel operation units, its processing ability is up to 4800MIPS. For multimedia applications, DM642EVM integrates 3 configurable video ports, McASP for audio applications, 10/100Mb/s Ethernet MAC and so on.
On the realization of the system, this system use CCS, adopt DSP/BIOS real-time OS, BSL (Board supported library), DDK (Driver develop kit), RF5 (Reference framework 5), CSL (Chip supported library) and other supported module to develop the software. Using of these module improving the programme efficiency, shorten the system implement period. And the system using C language to design and realize the audio communication system.
This system realized the point to point and point to multi-point audio communication system, each endpoint of the system is based on the DM642 platform. Each endpoint is consisted of two parts, signaling module and media module. Signaling module includes three tasks, signaling send and receives task and uart task. Media module including six tasks, audio capture task, audio coder and decoder task, audio send and receive task, and audio player task. Tasks of signaling module are synchronized by extern variable; Tasks of media module are synchronized using SCOM mechanism.
SCOM mechanism is a kind of synchonize mechanism which is offered by RF5. It build absolute message unilateralism channel for every task which need to synchronize with other tasks to realize the synchronize control, and use the message structure to realize the share and mutex of the resource. The disposal of the audio data between tasks in this system uses SCOM to realize synchronization. This system is to develop the software on the hardware platform, to accomplish the design and realization of nine modules talked above. The system is devided into two parts, signaling module and media module.
The signaling module is consisting of keyboard and UART. In this system, using the keyboard to simulate dial to trigger signaling, to achieve call setup, media negotiation and call realease and so on.
The media module is consisting of audio codec, DM642 and ethernet chips. After the setup of the signaling, the media modules achieve audio capture, process, and transport and so on. The principle of the system is this, firstly,control the AIC23 audio coder and decoder chip and audio port to realize the audio capture, keep the audio data captured in the capture buffer and compress the captured audio data using G.729A algorithm. Secondly, use the SCOM to transport audio data between tasks and use ethernet to complete the audio communication between DM642 and PC or DM642 and DM642. Finally, the receive endpoint play the audio after a series of reverse operation to realize the IP phone system based on the embed system.
At last, this pater tests the performance of the system and proposes related methods to improve the Qos performance of the system. First, the buffer mechanism of send and receive tasks, Second, audio coder and decoder algorithm. Then study these two methods separately and give the test result. This system adopts G.729A audio coder and decoder algorithm which is popular today. The algorithm can accomplish the speech compression in very low bit rate, suit for the demand of the IP phone today. In this system we wrapper the algorithm in a CELL according to the RF5 and optimize it according to the characters of the hardware platform. This paper also compare the former and after performance and prove that the performance after optimize is better than before.
According to what we have talked above, we can concluded that this article researches some key technologies about the audio communication based on embed system, designs and achieves a audio communication system and point to point and point to multi-point control signaling based on DM642 hardware platform.However realized the improvement on the problems exist in the system and proved that the effect is very well. For the sake of the time, this system is not perfect, and the author will improve it during afterward works.
引文
[1]黄永峰编著.因特网语音通信技术及其应用.人民邮电出版社.2002
[2]中国通信学会编著.IP 网络电话技术.人民邮电出版社,2000
[3]糜正琨.IP 网络电话技术.北京:人民邮电出版社.2000.6,17-19
[4]Specifications for the Network Voice Protocol (NVP), IETF, RFC741, 1977.
[5]W.Jiang, H.Schulzrine, Qos Evaluation of VoIP End-points, IEEE International Conference on Communications (ICC), Anchorage, Alaska, May 2003.
[6]ITU-T Recommendation H.323, Packet based multimedia communications systems. Feb 1998.
[7]RFC3261-Session Initiation Protocol 2002-6.
[8]RFC2543-SIP: Session Initiation Protocol 1999-3.
[9]IETF RFC 3611.“RTP Control Protocol Extended Reports(RTCP-XR)”, Nov 2005. http://www.ietf.org/rfc/rfc3611.txt
[10]Park Nam-Sup, Hwang Chong-Sun.Effective I/O scheme based on RTP for multimedia communication systems. Source:Journal of Computer Science and Technology, v 21, n 6, November, 2006, p989-996
[11]张登银.VoIP 业务 QoS 性能及其优化研究.江苏通信术,2005.1,6-9
[12]巴继东.IP 技术与综合宽带网.北京:北京邮电大学出版社,1999
[13]刘晓鸣.在MPLS网络中如何保证VOIP的QoS.武汉邮电科学研究院,2006
[14]Schimmel, Jiri Sysel, Petr.Control of digital audio signal processing over communication networks.Source: Lecture Notes in Computer Science, v 3421, n II, Networking - ICN 2005-4th International Conference on Networking, Proceedings, 2005, p425-432.
[15]Kim Eunsook,Kang Shin-Gak, “QoS support based on Intserv/DiffServ for SIP-based applications”, Managing Next Generation Convergence Networks and Services - 2004 IEEE/IFIP Network Operations and Management Symposium; Application Sessions, NOMS 2004.
[16]Salami, R., Laflamme, C., Bessette, B. and Adoul, J.P., 1997. ITU-T G.729Annex A: reduced complexity 8 kb/s CS-ACELP Codec for digital simultaneous voice and data. IEEE Commun. Mag., p53–63.
[17]Park, Eun-Chan; Choi, Chong-HoRobust delay estimator for playout buffering in Internet audio applications Source: Computer Communications, v 28, n 17, Oct 17, 2005, p1938-1946.
[18]刘璟.影响 IP 电话发展的关键问题研究.信息工程大学电子技术学院,2005
[19]Al-Shaer, Ehab; Brewster, Gregory B.On the impact of loss and delay variation on Internet packet audio transmission.Source: Computer Communications, v 29, n 10, Jun 19, 2006, p1578-1589
[20]Schuller, Gerald; Hirschfeld, Jens; Kraemer, Uli. Packet loss concealment in predictive audio coding.Source: IEEE Workshop on Applications of Signal Processing to Audio and Acoustics, 2005 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics, 2005, p227-230
[21]Bartkowiak, Maciej.Recent developments in audio coding.Source: IWSSIP 2005 - Proceedings of 12th International Workshop on Systems, Signals and Image Processing, IWSSIP 2005 - Proceedings of 12th International Workshop on Systems, Signals and Image Processing (SSIP-SPI, 2005), 2005, p45
[22]Al-Shaer, Ehab; Brewster, Gregory B.On the impact of loss and delay variation on Internet packet audio transmission.Source: Computer Communications, v 29, n 10, Jun 19, 2006, p1578-1589
[23]张雄伟,陈亮等编著.现代语音处理技术及应用.机械工业出版社,2003
[24]李长河,王永强.多媒体通信协议 H.324,H.323 及 SIP 的分析研究,微机发展,2003 年第 6 期,106-109
[25]IP 协议概述:http://www.jjtelecom.com.cn/txcp/2004312103228.htm
[26]S. Zeadally and F. Siddiqui, “Design and Implementation of a SIP-based VoIP Architecture”, Proceedings of the 18th International Conference on Advanced Information Networking and Application (AINA’04)
[27]艾明,“H.323 和 SIP 体系的比较和分析”.http://www.dtt.com.cn/tech/ techview.asp?id=145
[28]罗斯青.H323 和 SIP 的比较.通讯世界,2002-5
[29]TMS320DM642 Evaluation Module Technical Reference. August 2003.
[30]TMS320DM642 Video/Imaging Fixed-Point Digital Signal Processor (SPRS200G). August 2004.
[31]TMS320C6000 DSP Multichannel Audio Serial Port (McASP) Reference Guide Literature Number: SPRU041B May 2003
[32]TLV320AIC23 Stereo Audio CODEC 8 to 96kHz, With Integrated Headphone Amplifier. Data manual. 2001
[33]Texas Instruments Incorporated. TMS320 DSP/BIOS User’s Guide. November 2001, http://www.ti.com
[34]张宗橙.数字信号处理原理及应用.南京:东南大学出版社,1998.
[35]A DSP/BIOS AIC23 Codec Device Driver for the TMS320DM642EVM, Application Report SPRA922-June 2003
[36]DSP/BIOS Driver Developer's Guide Literature Number: SPRU616 November 2002
[37]Texas Instruments Incorporated. A Technical Overview of expressDSP-Compliant Algorithms for DSP Software Producers, 2002
[38]Texas Instruments Incorporated. TMS320 DSP Algorithm Standard, 2002
[39]TMS320C6000 TCP/IP Network Developer’s Kit Programmer’s Reference Guide. Jun. 2003.
[40]TMS320C6000 TCP/IP Network Developer’s Kit Technical Data Quick Reference Guide (SPRU568). October 2001.
[41]Synchronizing DSP/BIOS Threads (SPRAA25). Prashanth L A, Eldo Tony Kuruvilla. May 2004.5-17.
[42]Reference Frameworks for expressDSP Software: RF5, An Extensive, High-Density System. SPRA795A. TI application report. April 2003. Todd Mullanix, Davor Magdic, Vincent Wan, Bruce Lee, Brian Cruickshank, Alan Campbell, Yvonne DeGraw (technical writer). 40-43.
[43]ITU-T Recommendation G.729-Coding of Speech at 8kbit/s using Conjugate-Structure Algebraic-Code-Excited Linear-Prediction (CS-ACELP), 1996.
[44]Redwan Salami, Claude Laflamme. Design and description of CS-ACELP:a toll quality 8 kb/s speech coder. IEEE Transactions on Speech and Audio Processing, 1998, 6(2):116-130.
[45]唐昆.高质量 4~8kbit/s 变速率有限状态 ACELP 语音编码算法研究.电子学报,2000,28(1):21-25.
[46]MSDN Library, Visual C++ Documentation, Reference, “Microsoft Foundation Classes Library and Templates”, 1998
[47]Cisco Systems: "Understanding Delay in Packet Voice Networks" http://www.cisco.com/warp/public/788/voip/delaydetails.pdf, Document ID 5125,
[48]Salami, R., Laflamme, C., Bessette, B., Adoul, J-P., 1997b. Description of ITU-T recommendation G.729 Annex A: reduced complexity 8 kbit/s CS-ACELP Codec. In: IEEE Proceedings of International Conference Acoustic Speech Signal Processing, Vol. 2. p 775–778
[49]Jean-Emmanuel Chapron Bruno Charras. “An analysis of the IN call model suitability in the context of audio communication systems”, Computer Network 35,2001
[50]A.Markopoulou, F.Tobagi, and M.Karam, Assessment of audio communication systems quality over Internet Backbones, IEEE Transaction on Networking, October 2003
[2]中国通信学会编著.IP 网络电话技术.人民邮电出版社,2000
[3]糜正琨.IP 网络电话技术.北京:人民邮电出版社.2000.6,17-19
[4]Specifications for the Network Voice Protocol (NVP), IETF, RFC741, 1977.
[5]W.Jiang, H.Schulzrine, Qos Evaluation of VoIP End-points, IEEE International Conference on Communications (ICC), Anchorage, Alaska, May 2003.
[6]ITU-T Recommendation H.323, Packet based multimedia communications systems. Feb 1998.
[7]RFC3261-Session Initiation Protocol 2002-6.
[8]RFC2543-SIP: Session Initiation Protocol 1999-3.
[9]IETF RFC 3611.“RTP Control Protocol Extended Reports(RTCP-XR)”, Nov 2005. http://www.ietf.org/rfc/rfc3611.txt
[10]Park Nam-Sup, Hwang Chong-Sun.Effective I/O scheme based on RTP for multimedia communication systems. Source:Journal of Computer Science and Technology, v 21, n 6, November, 2006, p989-996
[11]张登银.VoIP 业务 QoS 性能及其优化研究.江苏通信术,2005.1,6-9
[12]巴继东.IP 技术与综合宽带网.北京:北京邮电大学出版社,1999
[13]刘晓鸣.在MPLS网络中如何保证VOIP的QoS.武汉邮电科学研究院,2006
[14]Schimmel, Jiri Sysel, Petr.Control of digital audio signal processing over communication networks.Source: Lecture Notes in Computer Science, v 3421, n II, Networking - ICN 2005-4th International Conference on Networking, Proceedings, 2005, p425-432.
[15]Kim Eunsook,Kang Shin-Gak, “QoS support based on Intserv/DiffServ for SIP-based applications”, Managing Next Generation Convergence Networks and Services - 2004 IEEE/IFIP Network Operations and Management Symposium; Application Sessions, NOMS 2004.
[16]Salami, R., Laflamme, C., Bessette, B. and Adoul, J.P., 1997. ITU-T G.729Annex A: reduced complexity 8 kb/s CS-ACELP Codec for digital simultaneous voice and data. IEEE Commun. Mag., p53–63.
[17]Park, Eun-Chan; Choi, Chong-HoRobust delay estimator for playout buffering in Internet audio applications Source: Computer Communications, v 28, n 17, Oct 17, 2005, p1938-1946.
[18]刘璟.影响 IP 电话发展的关键问题研究.信息工程大学电子技术学院,2005
[19]Al-Shaer, Ehab; Brewster, Gregory B.On the impact of loss and delay variation on Internet packet audio transmission.Source: Computer Communications, v 29, n 10, Jun 19, 2006, p1578-1589
[20]Schuller, Gerald; Hirschfeld, Jens; Kraemer, Uli. Packet loss concealment in predictive audio coding.Source: IEEE Workshop on Applications of Signal Processing to Audio and Acoustics, 2005 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics, 2005, p227-230
[21]Bartkowiak, Maciej.Recent developments in audio coding.Source: IWSSIP 2005 - Proceedings of 12th International Workshop on Systems, Signals and Image Processing, IWSSIP 2005 - Proceedings of 12th International Workshop on Systems, Signals and Image Processing (SSIP-SPI, 2005), 2005, p45
[22]Al-Shaer, Ehab; Brewster, Gregory B.On the impact of loss and delay variation on Internet packet audio transmission.Source: Computer Communications, v 29, n 10, Jun 19, 2006, p1578-1589
[23]张雄伟,陈亮等编著.现代语音处理技术及应用.机械工业出版社,2003
[24]李长河,王永强.多媒体通信协议 H.324,H.323 及 SIP 的分析研究,微机发展,2003 年第 6 期,106-109
[25]IP 协议概述:http://www.jjtelecom.com.cn/txcp/2004312103228.htm
[26]S. Zeadally and F. Siddiqui, “Design and Implementation of a SIP-based VoIP Architecture”, Proceedings of the 18th International Conference on Advanced Information Networking and Application (AINA’04)
[27]艾明,“H.323 和 SIP 体系的比较和分析”.http://www.dtt.com.cn/tech/ techview.asp?id=145
[28]罗斯青.H323 和 SIP 的比较.通讯世界,2002-5
[29]TMS320DM642 Evaluation Module Technical Reference. August 2003.
[30]TMS320DM642 Video/Imaging Fixed-Point Digital Signal Processor (SPRS200G). August 2004.
[31]TMS320C6000 DSP Multichannel Audio Serial Port (McASP) Reference Guide Literature Number: SPRU041B May 2003
[32]TLV320AIC23 Stereo Audio CODEC 8 to 96kHz, With Integrated Headphone Amplifier. Data manual. 2001
[33]Texas Instruments Incorporated. TMS320 DSP/BIOS User’s Guide. November 2001, http://www.ti.com
[34]张宗橙.数字信号处理原理及应用.南京:东南大学出版社,1998.
[35]A DSP/BIOS AIC23 Codec Device Driver for the TMS320DM642EVM, Application Report SPRA922-June 2003
[36]DSP/BIOS Driver Developer's Guide Literature Number: SPRU616 November 2002
[37]Texas Instruments Incorporated. A Technical Overview of expressDSP-Compliant Algorithms for DSP Software Producers, 2002
[38]Texas Instruments Incorporated. TMS320 DSP Algorithm Standard, 2002
[39]TMS320C6000 TCP/IP Network Developer’s Kit Programmer’s Reference Guide. Jun. 2003.
[40]TMS320C6000 TCP/IP Network Developer’s Kit Technical Data Quick Reference Guide (SPRU568). October 2001.
[41]Synchronizing DSP/BIOS Threads (SPRAA25). Prashanth L A, Eldo Tony Kuruvilla. May 2004.5-17.
[42]Reference Frameworks for expressDSP Software: RF5, An Extensive, High-Density System. SPRA795A. TI application report. April 2003. Todd Mullanix, Davor Magdic, Vincent Wan, Bruce Lee, Brian Cruickshank, Alan Campbell, Yvonne DeGraw (technical writer). 40-43.
[43]ITU-T Recommendation G.729-Coding of Speech at 8kbit/s using Conjugate-Structure Algebraic-Code-Excited Linear-Prediction (CS-ACELP), 1996.
[44]Redwan Salami, Claude Laflamme. Design and description of CS-ACELP:a toll quality 8 kb/s speech coder. IEEE Transactions on Speech and Audio Processing, 1998, 6(2):116-130.
[45]唐昆.高质量 4~8kbit/s 变速率有限状态 ACELP 语音编码算法研究.电子学报,2000,28(1):21-25.
[46]MSDN Library, Visual C++ Documentation, Reference, “Microsoft Foundation Classes Library and Templates”, 1998
[47]Cisco Systems: "Understanding Delay in Packet Voice Networks" http://www.cisco.com/warp/public/788/voip/delaydetails.pdf, Document ID 5125,
[48]Salami, R., Laflamme, C., Bessette, B., Adoul, J-P., 1997b. Description of ITU-T recommendation G.729 Annex A: reduced complexity 8 kbit/s CS-ACELP Codec. In: IEEE Proceedings of International Conference Acoustic Speech Signal Processing, Vol. 2. p 775–778
[49]Jean-Emmanuel Chapron Bruno Charras. “An analysis of the IN call model suitability in the context of audio communication systems”, Computer Network 35,2001
[50]A.Markopoulou, F.Tobagi, and M.Karam, Assessment of audio communication systems quality over Internet Backbones, IEEE Transaction on Networking, October 2003