IP Bearer Network for NGN

1 Requirements on Bearer Network for NGN (Softswitch)
Worldwide NGN markets show that NGN is deployed not only on national backbone network, but also at core, convergence and access layers of Metropolitan Area Network (MAN). For example, China Telecom began to build the second IP backbone last year, known as CN2. One of the main services provided by the CN2 project is the trunk circuits, featuring light load mode to ensure QoS.

    To bear NGN (Softswitch) services with high efficiency, the requirements for IP bearer network may be analyzed from the following aspects:

• Structure requirements
• Bandwidth requirements
• Manageable and controllable engineering administration
• Network security
• Supporting IPv6
• Virus prevention, detection and measurement mechanisms
• Remote disaster tolerance solution
  

    In one word, the NGN requires all-IP products such as routers and Ethernet Switches (ES) to achieve high efficiency and reliability, as well as advanced network technologies. The required technologies are QoS, Multiple Protocol Label Switching (MPLS), MPLS Virtual Private Network (VPN), MPLS-TE (Traffic Engineering) and IPv6 to ensure QoS, security, reliability and more.

2 Comparison of Bearer Modes
Currently, ZTE´s equipment can support three kinds of bearer modes or their combination to bear commercial Softswitch services for international or domestic telecom operators: Fiber Connection (FC), existing MAN and
Multi-Service Transmission Platform (MSTP).

    Table 1 makes the comparison from perspectives of QoS, security, resource utilization, and more.
ZTE presents its thoughts for NGN bearer network on the basis of comprehensive considerations of the above comparison results, pressing deployment of Softswitch and high requirements for bearer network (e.g. QoS, security, reliability, fast fault location and quick repair & maintenance).

    （1）Use MPLS VPN to construct NGN virtual private bearer network so as to logically separate NGN services from traditional data services. Therefore, the security of NGN bearer network is guaranteed.

    （2）Adopt advanced technology such as light load, fast route convergence, DiffServ (Differentiated Services), TE (Traffic Engineering) and FRR (Fast Reroute) to ensure QoS.

3 QoS Requirements of Voice over IP Network  (VoIP)

3.1 Time Delay
For a VoIP network, voice quality is directly affected by three QoS quality factors: delay, jitter and packet loss. In fact, delay and relevant echo will greatly degrade the understandability of the conversation. This phenomenon is especially obvious in a VOIP network.

    The delay together affects the quality of voice so greatly that echo interference and interactivity would be worsen.
In general, the end-to-end delay may be divided into two parts: fixed delay and variable delay.

    Fixed delay consists of codec delay and packing delay. It is determined by compression algorithm and the amount of voice being packed. Even though it is unlikely to optimize the fixed delay, it can still be improved by selecting appropriate compression algorithm, reducing the amount of voice packed, allocating reasonable Digital Signal Processing (DSP) load and adopting good pipeline processing flow.

    Variable delay consists of transfer delay on the bearer network, queuing delay at nodes, service processing delay and de-jittering delay, which are nearly determined by the port rate, network load, passed route, QoS mode and QoS algorithm. In particular, de-jittering delay is correlative to the index of jitter of a bearer network. When voice signals pass different networks, jitter may be generated. However, it is possible to greatly reduce the jitter by carefully adopting suitable network technology. Consequently, the jitter delay may be reduced as well.

    As long as the dynamic switching time for audio encoding is less than 60 ms, and IP network delay is less than 80 ms, there will be no chopped or jittered audio.

    (1) Network condition is good: average (Perceptual Speech Quality Measurement) PSQM is <1.5.

    (2) Network condition is bad (packet loss rate =1%, network jitter =20 ms, delay =100 ms): average PSQM is <1.8.

    (3) The worst environment (packet loss rate =5%, network jitter =60 ms, delay =400 ms): average PSQM is <2.0.
    Generally speaking, the delay (loop-back) produced by Gateway (GW) equipment may be divided into codec delay, receiver input buffer delay and internal queuing delay.

• Delay for G.729 is <150 ms
• Delay for G.723.1 is < 200 ms
• Delay for G.711 is < 120 ms
  

    When the end-to-end one way delay is < 150 ms, it is acceptable for most applications. When the delay is in the range of 150 ms to 400 ms, it is acceptable, provided that users are aware of the delay condition in advance. When the delay is > 400 ms, it is unacceptable.

    Therefore, IP bearer network must meet the following conditions:
    (1) The larger the number of hop, the longer delay will be. It is recommended the end-to-end hop count be less than 16. If possible, make the hop count as small as possible.

    (2) Transfer delay for each hop is <10 ms.

    (3) Basic requirements for end-to-end delay is <400 ms in the worst situation, <150 ms in general situation.
While considering GW delay (ring-back), the requirements on IP bearer network are recommended
as below:

• G.729 delay is <250 ms in the worst situation, <75 ms in general situation;
• G.723 delay is < 200 ms in the worst situation, <50 ms in general situation;
• G.711 delay is <280 ms in the worst situation, <90 ms in general situation.

3.2 Jitter
Jitter stems from variability in packet delay. Consequently, it may cause end-to-end delay to be increased and quality of voice to be degraded.

    Jitter is due to network congestion. Since both voice and data are transmitted on the same physical line, it is common that voice and data may be blocked because the same physical line is occupied by data packet.

    Assuming that G.711 is adopted in a LAN, if only to assess the impact of jitter, the jitter influence on quality of voice is described in Table 2.

    The following measures are often taken to solve the jitter problem:
    (1) Jitter buffer queue: devices such as GW and Integrated Access Device (IAD) have JitterBuff queue to be used for jitter cancellation

    (2) QoS policy on IP bearer network: make sure voice data has the highest priority by making appropriate QoS policy. The main method to solve the jitter problem is to allocate sufficient bandwidth for voice data and let the voice data be sent firstly.

3.3 Packet Loss Rate
Packet loss has considerable negative impact on perceived voice quality for a VoIP network. When G.711 is adopted in a LAN, if only to assess and analyze the impact of packet loss, the relation between packet loss rate and voice quality is illustrated in Table 3.

    From Table 3, we know that voice quality is unacceptable if packet loss rate is > 10%. The minimum requirement for packet loss rate is 5% if the voice quality is acceptable. Therefore, the packet loss rate for an IP bearer network must be less than 5%.

3.4 Network Bandwidth
Sufficient bandwidth is essential for QoS. For example, an IP call (G.729 20 ms) with clear voice needs about 32 kb/s bandwidth so it is impossible to carry four channels of this IP calls over a 64 kb/s link. Additionally, if the bandwidth of IP backbone network is wider, the total transfer delay for data passing through the whole network will be decreased.

    In one word, for VoIP to work well and QoS to be assured, network links must be properly sized with sufficient bandwidth for real-time voice traffic.

4 Networking Case Study
Figure 1 shows one of ZTE´s networking solutions. This network is constructed with about 80 devices. Those include ZXR10 T128 and ZXR10 T64E (telecom-grade high-end routers), ZXR10 GER (telecom-grade General high-Efficient Router), ZXR10 T16C (intelligent C-family MPLS routing switch), ZXR10 3904 (routing switch) and ZXR10 2826/2826E/2618/1816 (layer-2 Ethernet switch).The IP bearer network features high and reliable performance, high security, wide bandwidth, and guaranteed QoS.

    Both ZXR10 T128 and ZXR10 T64E adopt telecom-grade redundancy configuration in modules and components such as power supply, switch and main control.  Virtual Router Redundancy Protocol (VRRP) is supported for communication between ZTR10 T128 and ZXR10 T64E, and between two ZXR10 3904 router switches.
To date, the following services are successfully deployed:

• Basic and supplementary voice services with 100 000 subscribers
• WAC service
• Universal Personal Telecommunication (UPT) service
• Point-to-point audiovisual service
• WEB 800 service
• UC service
• Click to Dial (CTD) service (web call service)
• IP CENTREX
• Diversified value-added services such as software terminal and hardware terminal services
• Interconnection and interworking among Softswitch equipment from different vendors or geographically separated Softswitch systems
• 179xxIP phone interoperability test
• One-number-express + color ring service, and more
  

5 Conclusion
ZTE´s solutions can optimize and reconstruct the current IP MAN, or build a virtual private IP network by adopting
"FC+MSTP" mixed mode. It features high efficiency, security and controllable engineering. Moreover, these solutions support the transition to IPv6.

Manuscript received: 2005-06-05