资源描述
,IMS,2G/3G,VoLTE网络架构,LTE,eNodeB,SGW,HLR/ HSS,IMS-HSS,HLR,LTE-HSS,PGW,MME,PCRF,GERAN/UTRAN,MSS,SGSN,MGW,P-CSCF,A-BGW,I-CSCF,S-CSCF,TAS,UE,MRFC,MRFP,IMS-MGW,MGCF,BGCF,MRFP,VoLTE网元列表,VoLTE网元列表,Need to be added 3GPP,VoLTE Network Interface,VoLTE bearer concept,Voice traffic - QCI 1,SIP Signaling - QCI 5,Video traffic QCI 2,EPC,eNB,VoLTE requires own IMS APN: SIP signaling always uses a QCI = 5 bearer with non-GBR Voice traffic always uses a dedicated QCI =1 bearer with GBR Other IMS media - if present, e.g. for RCS - use QCI = 8/9 media with non GBR Internet traffic for another (internet) APN,UM DRB,UM DRB,AM DRB,IMS APN,Best effort data - QCI 9,AM DRB,Internet APN,other IMS media - QCI 8/9,AM DRB,QoS aspects of E2E VoLTE Call Flow (IR.92),Thomas Hammann,2e,Precondition,Ulf Spangenberger,Precondition: a set of constraints that must be fulfilled before the user is alerted (rfc3312) new SDP attributes curr (current status), des (desired status), conf (confirmation status) Offer/Answer message flow,Initial INVITE (SDP),183 Session Progress (SDP),UPDATE (SDP),200 OK (SDP),RAB setup A,RAB setup B,180 Ringing,A - B Delay ringing until curr=des for A,A,B,B - A Inform me when curr=conf,A - B curr=conf=des,IMS Invite Procedure Rx Interface,Ulf Spangenberger,2e,P-CSCF A,S-CSCF A,S-CSCF B,P-CSCF B,PCRF,PCRF,INVITE (SDP) IP address-A, port-A, ordered list of supported codecs, bandwidth,180 RINGING,200 OK (SDP) IP address-B, port-B, ordered list of negociated codecs, bandwidth,ACK,SAE-GW,SAE-GW,Setup dedicated bearer with QCI=1,Access Authorization, AAR/A,Flow descriptor: UL/DL for RTP/RTCP Bandwidth: UL/DL for RTP/RTCP Codec list,Call routing (IMPU),UE-A,UE-B,PCRF for VoLTE,2e,Elisabeth Koziel,PGW,PCRF,Create Dedicated Bearer Req,PCC Rules Provisioning (PCC Rules, Event Triggers),Gx RAA,Create Dedicated Bearer Resp,Services,P-CSCF,Session and media info,PGW decides based on the QCI/ARP pair whether an existing bearer can be used or whether a new dedicated bearer is needed for VoIP,1,2,3,4,5,Default Bearer for signaling created,0,Gx_CCR-I, QCI=5,ARP=2,SessionId=1,Rx_AAR:MediaComponenDescription,Gx_RAR:Charging-Rule-Install(Flow,Qos) ,SessionId=1,Rx AAA,IR.92 based QoS model: IMS APN is used for the IMS traffic, other APN for data traffic,E2E VoLTE Call Flow (IR.92),2e,Phase 1 Attach / Primary Default Bearer with IMS APN (QCI=5),Phase 2 SIP Register w/ primary Default Bearer / 2nd PDN connection for Data APN,Phase 3 Initiate VoIP call using Primary default bearer,Phase 4 Dedicated Bearer for VoIP Data,Phase 5 User Data and GBR/Non-GBR sheduleing,Phase 6 Release of VoIP call and dedicated bearer,LTE/EPC Point of View,Phase 1 SIP Register,Phase 2 SIP Invite,Phase 3 SIP Release,IMS Point of View,Option 1 Combined EPS/IMSI combined EPS/IMSI attach,UE behaviour for IMS PS Voice preferred with CS Voice as secondary TS23.221,Option 2 non combined EPS/IMSI attach,X,VoLTE呼叫流程,VoLTE Handover Procedures SRVCC Handover for Voice from LTE to 2G/3G CS,IMS/VoIP/VoLTE Domain,CS coverage,LTE/EPS Domain,2G/3G CS & PS Domain,S6a,D,UE (A-Party),Gr /S6d,UE (A-Party),LTE coverage,Gm,Cx,ISC,Sv / SGs,Gx,Rx,Mw,Single Radio Voice Call Continuity (SR-VCC),eNB,SAE-GW,MME,PCRF,CSCF (of A-Party),HSS / HLR,MSC-S,MGW,TAS,BSC/RNC,SGSN,GGSN,Other Operators,LTE Cell,CSCF (of B-Party),SCC-AS,UE (B-Party),S1-U,S1-MME,S11,Voice path when the A-Party is in LTE,Voice path when the A-Party is in CS,SRVCC related signalling,Signalling interfaces,MAP,SIP Invite (STN-SR, SDP=MGW),0. User plane before SR-VCC Handover,6. User plane after SR-VCC handover,4. The SCC AS initiates a SIP Re-Invite to the B-Party to switch the User Plane from PDN-GW to the MGW,5. Triggered by the SIP Re-Invite, the B-party switches the User Plane to the MGW,2. MME triggers “PS to CS Hand-over Request” to MSC-S including Target Cell ID, STN-SR, etc.,1. eNb initiates SR-VCC Handover,3. The MSC-S reserves CS radio resources and initiates a SIP Invite towards the IMS (SCC-AS) using the STN-SR.,STN-SR Session Transfer Number for SRVCC (static configured in HSS) - SCC-AS,IMS/VoIP/VoLTE Domain,CS coverage,LTE/EPS Domain,2G/3G CS & PS Domain,S6a,D,UE (A-Party),Gr /S6d,UE (A-Party),LTE coverage,Gm,Cx,ISC,Sv / SGs,Gx,Rx,Mw,Single Radio Voice Call Continuity (SR-VCC),Sh,eNB,SAE-GW,MME,PCRF,CSCF (of A-Party),HSS / HLR,MSC-S,MGW,TAS,BSC/RNC,SGSN,GGSN,Other Operators,LTE Cell,CSCF (of B-Party),SCC-AS,UE (B-Party),S1-U,S1-MME,S11,Voice path when the A-Party is in LTE,Voice path when the A-Party is in CS,SRVCC related signalling,Signalling interfaces,MAP,ATGW,ATCF,MGW,P-GW,0. User plane is before SR-VCC Handover,2. MME triggers “PS to CS Hand-over Request” to MSC-S including Target Cell ID, STN-SR, etc.,1. eNb initiates SR-VCC Handover,6. User plane after SR-VCC handover,3. The MSC-S reserves CS radio resources and initiates a SIP Invite towards the ATCF using the STN-SR.,5. SCC-AS is informed for the SRVCC HO without the bearer change indication. Since there is no SDP update, no SIP Re-Invite is sent to the B-Party.,4. ATCF configures ATGW with SDP=MGW which switches the bearer from PDN-GW (LTE) to the MGW (CS),VoLTE Handover Procedures eSRVCC Handover for Voice from LTE to 2G/3G CS,STN-SR Session Transfer Number for SRVCC (updated during Registration) - ATCF,AMR简介,全称Adaptive Multi-Rate，自适应多速率编码，主要用于移动设备的音频，压缩比比较大，但相对其他的压缩格式质量比较差，由于多用于人声，通话，效果还是很不错的。,1. AMR: 又称为AMR-NB，相对于下面的WB而言， 语音带宽范围：3003400Hz， 8KHz抽样 2. AMR-WB:AMR WideBand， 语音带宽范围： 507000Hz 16KHz抽样 “AMR-WB”全称为“Adaptive Multi-rate - Wideband”，即“自适应多速率宽带编码”，采样频率为16kHz，是一种同时被国际标准化组织ITU-T和3GPP采用的宽带语音编码标准，也称 为G722.2标准。AMR-WB提供语音带宽范围达到507000Hz，用户可主观感受到话音比以前更加自然、舒适和易于分辨。,AMR-NB,AMR 一共有16种编码方式， 0-7对应8种不同的编码方式， 8-15 用于噪音或者保留用。,AMR-WB,时延和抖动,时延：端到端包传递的时间,ITU G.114规范建议，在传输语音流量时,单向语音包端到端延迟要低于150ms(对于国际长途呼叫,特别是卫星传输时,可接受的单向延迟为300ms。如果超过300ms则通话的质量会变的让人不能忍受。过多的包延迟可以引起通话声音不清晰、不连贯或破碎。,经验：大多数用户察觉不到小于100毫秒的延迟，当延迟在100毫秒和300毫秒之间时，说话者可以察觉到对方回复的轻微停顿。这种停顿可以影响到通话双方的交流。超过300毫秒，延迟就会很明显，用户开始互相等待对方的回复，通话过程变成类似对讲机式的模式。而且较长的时延也会将回声问题的影响放大。,时延的产生因素,编码的处理：模拟形式的声音信号在CODEC被采样和量化为PCM信号，DSP对PCM信号进行压缩处理所产生的时延为编码处理时延。 这种时延产生在设备侧，如果设备的编码器固定，则编码时延也固定。 包化：包化就是将编码器输出的语音净荷放置到RTP/UDP/IP包中的过程，相对于编码的时延，包化的时延很小，因为包化的过程没有复杂的运算，仅仅是增加包头和计算校验和，而编码则有大量的数学运算。 队列（Queuing）：语音的净荷放置到IP包中后，要被设备转发到目的地，这些包会在设备的出接口队列中，等待被调度。转发设备不同的队列机制对IP包的处理有很大不同。可以通过合理的配置来减少语音包在队列中等待的时间，进而减少队列时延。 串行化（Serialization）：接口队列中的语音IP包，被送离设备前会放置到接口的物理队列当中，如果物理队列中有一个较大分组，还在发送状态，则语音分组必须等待这个较大的分组发送完毕后才能发送，这个等待的时间就是串行化时延。比如一个时钟速率为64kbps的链路要发送一个1600Bytes大小的FTP分组，则串行化产生的时延会达到200ms（16008/640001000）。这对于后面等待的语音包来说已经是很大的时延了。 广域网时延：对于ISP提供的广域网链路，对于用户来说只是一个黑盒子，除了上述的编码时延外，构成广域网链路的路由器交换机都会产生包化、队列、串行化的时延。而且到达同一目的的路径不同，其每个包的时延也不同，而这些时延对于用户来说是不可控的，当然我们在租用ISP的线路时，可以要求ISP提供符合时延要求的线路。,抖动,变化的时延被称作抖动（Jitter），抖动大多起源于网络中的队列或缓冲，尤其是在低速链路时。而且抖动的产生是随机的，比如你无法预测在语音包前的数据包的大小，既便你使用LLQ，如果大数据包正在传输过程中，当语音分组到达时，它还是要等待数据分组被发送完。而在低速的链路中，语音数据混传时，抖动是不可避免的。通常使用LFI将大包拆小，来减少大包对时延的影响。,MOS,MOS （Mean Opinion Score）指主观评判得分 MOS分值与听觉感受的对应关系,谢谢！,
展开阅读全文