Choosing the Right Audio Codecs for VoIP over cdma2000 Networks:

Transcription

1 Choosing the Right Audio Codecs for VoIP over cdma2000 Networks: System capacity, Voice quality, Delay, and Transcoding issues Dr. Sassan Ahmadi NOKIA Inc. February 8, CDG VoIP Summit San Diego, CA/February 2005/Sassan Ahmadi

2 Outline Drivers and advantages of VoIP over cdma2000 networks Requirements for VoIP over cdma2000 networks Delay requirement Mobile-to-mobile delay, an example 3GPP2 standard codecs for VoIP System capacity Simulation parameters Simulation results Migration path to VoIP over cdma2000 networks Summary Appendix Definitions CDG VoIP Summit San Diego, CA/February 2005/Sassan Ahmadi

3 Drivers and Advantages of VoIP over cdma2000 Networks VoIP drivers Efficient use of system bandwidth as well as flexible technology. The use of standard packet network components and protocols with lower costs. Migration toward all-ip network architecture in general and adoption of 1x EV-DO by cdma operators as the next cdma air interface in particular. VoIP is the simplest and most efficient path to convergence of the Next Generation Networks. VoIP facilitates the introduction of rich multimedia services. Integration of voice, data, and other multimedia services simplifies network design, operation, and management. Convergence of various cellular/wireless/fixed network services CDG VoIP Summit San Diego, CA/February 2005/Sassan Ahmadi

4 Drivers and Advantages of VoIP over cdma2000 Networks Advantages of VoIP over CS voice services More flexible and efficient codec design The permissible encoding rates in existing cdma2000 codecs are restricted by the air-interface to either rate-set I or rate-set II. Discontinuous transmission cannot be used in cdma2000 CS voice services. Opportunity for cross-system interoperability. Opportunity for providing better voice quality. Existing cdma2000 CS voice services involve transcoding VoIP can offer transparent end-to-end transport, resulting in Lower end-to-end delay by eliminating transcoding Improving quality of service by avoiding quality degradation due to transcoding Reducing the complexity and cost of the gateways Ease of harmonization with other cellular/fixed network voice services CDG VoIP Summit San Diego, CA/February 2005/Sassan Ahmadi

5 Requirements for VoIP over cdma2000 Networks Match or exceed the quality of existing CS voice services. One-way delay equivalent to CS voice services is desirable. Match or exceed the capacity of CS voice over cdma2000 1x. Interoperability with other cellular and fixed networks to avoid transcoding is desirable. The evolutionary path should include an overlap period with the existing CS voice services to allow for smooth transition. Reasonable robustness and tolerance with respect to packet loss and jitter in the IP networks. Use and enhancement of IMS architecture to control and manage SIP/SDP based sessions and signaling. Some enhancements in the mobility management to overcome the limitations on the real-time services such as VoIP (e.g., handoffs). Compliance with regulatory services and requirements CDG VoIP Summit San Diego, CA/February 2005/Sassan Ahmadi

6 Delay Requirement ITU-T E-Model 100 Users very satisfied 90 E-model rating R Users satisfied Some users dissatisfied Many users dissatisfied Maximum Perceptually Tolerable Delay Mouth-to-ear-delay/ms Nearly all users dissatisfied G.114_F01 ITU- G.114 Determination of the effects of absolute one-way delay by the E-model CDG VoIP Summit San Diego, CA/February 2005/Sassan Ahmadi

7 Mobile-to-Mobile Delay An Example Encoder Encoder 1xEV-DO 1xEV-DO Modem Modem 1xEV-DO 1xEV-DO RAN RAN + PDSN PDSN 35 ms 5 ms RL H-ARQ Delay 60 ms 20 ms IP Network Included in capacity calculations 15 ms Decoder Decoder De-Jitter De-Jitter Buffer Buffer 1xEV-DO 1xEV-DO Modem Modem 1xEV-DO 1xEV-DO RAN RAN + PDSN PDSN 20 ms ms 5 ms 20 ms FL H-ARQ Delay (voice packet drop timer) 80 ms Transcoding delay has not been included ~ 40 ms CDG VoIP Summit San Diego, CA/February 2005/Sassan Ahmadi

8 3GPP2 Standard Codecs for VoIP 3GPP2 Standard Codec Wideband Speech Processing Capability Narrowband Speech Processing Capability Maximum Source Bit Rate (kbps) Operating cdma2000 Rate-Set Algorithmic Delay (ms) DTX Capability Interoperability with UMTS Codecs Multimode Functionality (Quality- Capacity Tradeoff) QCELP Rate-Set II 30.5 EVRC 8.55 Rate-Set I 33 SMV 8.55 Rate-Set I 35.5 VMR-WB Rate-Set I Rate-Set II UMTS/AMR-WB VMR-WB is the only 3GPP2 standard codec with flexible DTX capability VMR-WB is the only 3GPP2 standard codec with both wideband and narrowband processing capabilities VMR-WB is the only 3GPP2 standard codec that is interoperable with UMTS/AMR-WB codec at 12.65, 8.85, and 6.6 kbps CDG VoIP Summit San Diego, CA/February 2005/Sassan Ahmadi

9 System Capacity Simulation Parameters (1xEV-DO Rev. A) Parameters Channel model Preamble detection error DRC channel error ACK/NACK channel error H-ARQ RLP Retransmission Control Channel Overhead Multi-user packet Voice packet drop timer Target FER Voice traffic model RTP/UDP/IP Header Size Value Channel B Modeled Error free Error free Modeled Disabled t Modeled Modeled 80 ms 1% Fixed inter-arrival time of 20 ms 4 Bytes Comment As specified in [1] Typically 1/16 of the total capacity is allocated Timer starts from the time each voice packet arrives in BS buffer Forward Link Inter-arrival time between voice packets Average compressed header size for IPv6 Channel Model Multi-path Model # of Fingers Speed (km/h) Fading B Pedestrian B 3 10 Jakes [1] "1xEV-DO Evaluation Methodology (V1.4)," C , 3GPP2 WG3 DO Ad Hoc Group, October 2, CDG VoIP Summit San Diego, CA/February 2005/Sassan Ahmadi

10 System Capacity Simulation Results Our simulation methodology uses those parameters specified in the strawman [1,2] except Channel model Channel model B only was studied as some concerns were raised regarding the percentage of channel model A users in mixed channel environment Outage criteria Based on long-term FER Requirement: Less than 3% of all users in the system have long-term FER higher than 1% EVRC codec was used for the cdma2000 1x simulation. Channel Model cdma2000 1x Circuit-switched w/o RX Diversity (users/sector) VoIP over 1x EV-DO Rev. A w/o RX Diversity (users/sector) EVRC [Benchmark] ( DTX) EVRC* [Benchmark] (with DTX) VMR-WB Mode 0 (with DTX) VMR-WB Mode 4 (with DTX) Channel B VMR-WB mode 4 offers superior wideband quality while performing statistically equivalent to EVRC in narrowband operation, operating at an average data rate that is 8% lower than that of EVRC. [1] "1xEV-DO Evaluation Methodology (V1.4)," C , 3GPP2 WG3 DO Ad Hoc Group, October 2, 2004 [2] "1xEV-DV Evaluation Methodology Addendum (V14)," 3GPP2 WG5 Evaluation Ad Hoc, June 16, 2003 * te: The existing EVRC standard codec does not include DTX capability and this number represents only an estimate, should a DTX mechanism similar to that of VMR-WB be used for EVRC CDG VoIP Summit San Diego, CA/February 2005/Sassan Ahmadi

11 Migration Path to VoIP over cdma2000 Networks cdma2000 1x CS voice services Next Generation Networks VoIP over cdma2000 Networks CS voice as the main service VOIP as an add-on service VOIP as the main service VoIP over Fixed Networks CDG VoIP Summit San Diego, CA/February 2005/Sassan Ahmadi

12 Summary VoIP over cdma2000 networks is promising and can take off as early as 2H 2006 as an add-on service. cdma2000 1x system capacity and voice service quality can be matched or exceeded with certain assumptions. The use of RX diversity with 1xEV-DO Rev. A is one way to further increase the capacity gains. Harmonization of VoIP services across various cellular/fixed networks can be facilitated through the use of interoperable codecs (e.g., UMTS/AMR-WB, CDMA/VMR-WB), resulting in more efficient network design and operation by eliminating the transcoding. The 3GPP2 standard wideband/narrowband multimode speech codec, VMR-WB, provides unique functionalities that are ESSENTIAL to the success of VoIP over cdma2000 networks. 3GPP2 VMR-WB standard offers superior quality, DTX capability, and interoperability with UMTS/AMR-WB, at a system capacity that is comparable to that offered by other existing 3GPP2 standard codecs CDG VoIP Summit San Diego, CA/February 2005/Sassan Ahmadi

13 Thank You CDG VoIP Summit San Diego, CA/February 2005/Sassan Ahmadi

14 Appendix Definitions Narrowband speech codec: A vocoder that operates with 8 khz sampled input/output speech signals that are band-limited to Hz. Wideband speech codec: A vocoder that operates with 16 khz sampled input/output speech signals that are band-limited to Hz. The expanded audio bandwidth results in more intelligibility and naturalness and substantially improved quality of the speech signals. Discontinuous transmission (DTX): A mechanism that suppresses the encoding and transmission of inactive (silence) intervals in a speech signal, thus reducing the output data rate of the vocoder. Silence speech intervals are perceptually less significant and therefore the impact on speech quality is minimal. Transcoding: Extra decoding and encoding functions at the gateways to enable connection of different codecs at the end points. This process adds to the overall delay and degrades the speech quality. Interoperability: Capability of a vocoder to decode the speech data encoded by another codec without incorporating any transcoding mechanism. Multimode speech coding: Tradeoff between voice quality and system capacity through the use of multiple encoding schemes each operating at different bit rate CDG VoIP Summit San Diego, CA/February 2005/Sassan Ahmadi