1 Introduction
NGN is the development trend of the next-generation telecommunications network. Although its system architecture and related standards are still being improved, the requirements for the bearer network technology are the same whether it is based on the softswitch architecture or the IMS-based architecture. All need bearer network to provide carrier-class QoS and security.
This paper analyzes the QoS and security issues of the bearer network, and gives a realistic NGN bearer network construction plan.
2NGN bearer network QoS considerations
Considering the QoS of the bearer network, we must first clarify several important factors that affect VoIP QoS.
2.1 Delay
Due to the inherent characteristics of the current IP packet network and the use of low-bit voice codecs, the end-to-end delay of VoIP voice packets is much larger than the delay in circuit-switched networks, and the components are more complicated. VoIP applications The diversity of the network communication structure and the underlying transmission protocols in China determines the diversity of delay components.
End-to-end delay can be divided into two parts, namely fixed delay and variable delay. The fixed delay includes the delay introduced by the codec and the packing delay. The fixed delay is related to the compression algorithm used and the amount of voice data packaged. Variable delay includes: transmission on the bearer network, queuing in the node, service processing delay, debounce delay, these and the device's port rate, network load, network path passed, equipment support for QoS, implementation The QoS algorithm is closely related. In particular, the debounce delay is closely related to the jitter index of the bearer network. Using appropriate network technology can significantly reduce the jitter introduced when voice passes through the network and reduce the debounce delay.
The end-to-end delay of voice packets in the IP network, the delay below 150ms is acceptable for most applications; the delay between 150 ～ 400ms is acceptable under the premise that the user predicts the delay status; Delays greater than 400ms are unacceptable.
Currently, network devices of different levels have a packet processing delay of tens of microseconds to several milliseconds under normal circumstances, which can meet the single-hop delay requirement, but the number of hops of the bearer network cannot exceed the above end-to-end time. Delay requirements, and the fewer hops, the better.
2.2 Jitter
According to the actual measurement, it is found that jitter greater than 500ms is unacceptable, and it is acceptable when the jitter reaches 300ms. At this time, in order to eliminate jitter will cause a large delay, the impact of comprehensive delay on voice quality is considered, and the bearer The jitter of the network is less than 80ms.
Jitter will cause an increase in end-to-end delay and will cause a decrease in voice quality. The factors that affect jitter are generally related to the degree of network congestion. The network node traffic is too busy, and the buffer time of data packets in each node is too long, which makes the arrival rate change greatly. Since voice and data are transmitted on the same physical line, voice packets are usually blocked due to the burstiness of the data packets.
2.3 Packet loss rate
Packet loss has a greater impact on VoIP voice quality. When the packet loss rate is greater than 10%, it is no longer acceptable, and when the packet loss rate is 5%, it is basically acceptable. Therefore, the packet loss rate of the IP bearer network is required to be less than 5%.
There are two main reasons for the formation of the packet loss rate. One is the error code in the traditional IP transmission process. This situation has a very low probability of occurring under the current network conditions. The other is caused by the inability to guarantee business bandwidth. The more congested the network traffic, the stronger the impact and the greater the rate of packet loss.
2.4 Bandwidth
Sufficient bandwidth is an important means to guarantee business QoS. For example, the voice coding compression adopts the ITU-TG.729 standard, and the rate is 8 kbit / s. A typical voice encoder distributes a voice data packet every 20 ms, and each data packet contains two voice frames, so every 20 ms a 160-bit signal is generated, that is, the data packet size is 20 bytes. After adding 12 bytes of RTP header, 8 bytes of UDP header and 20 bytes of lP header, the size of each packet becomes 60 bytes. Therefore, the effective rate of each voice connection is (60 & TImes; 8) / 20 = 24kbit / s. Similarly, the effective rate of G.711 is 80kbit / s. Consider that the main link layer technology in the current metropolitan area network is Ethernet. The effective rates of G.729 and G.711 are: 34.4kbit / s and 90.4kbit / s, respectively.
The detailed calculation principles of the bandwidth of the control flow and the signaling flow are the same. The number of bytes of the signaling message and overhead must be considered. The calculation of the overhead is similar to the calculation of the bandwidth of the voice service. Calculate the number of call messages in the protocol and the allocation of call proportions. Since the bandwidth occupied by control signaling in the bearer network is very small compared to media streams, it only accounts for about 0.5% of the bandwidth required by G.711 encoding. A simple and fast algorithm is to reserve 2.5% of the bandwidth of the media stream.
For any level of bearer network equipment, the upstream port aggregates NGN services, and its bandwidth design must meet the bandwidth requirements of NGN users carried by other ports and downstream equipment.
2.5 QoS considerations
Through the analysis of the indicators that affect QoS, it can be seen that the careful design of the bearer network (such as layer and hop control), adequate and reasonable bandwidth planning, and avoiding network congestion are factors that need to be considered in the current practical solution.
The current IP service quality architecture mainly includes the IntServ system and DiffServ system recommended by the IETF. The IntServ model uses the Resource Reservation Protocol (RSVP) to reserve network resources according to the service quality requirements of the service before transmitting data, thereby providing end-to-end service quality assurance for changing the data flow. Although the integrated model can provide a definite quality of service guarantee, it needs to maintain the state of each flow in the network. It has high requirements on routers and is difficult to implement in a large IP network. Therefore, this solution is not considered. The basic idea of ​​differentiated services is to classify user data streams according to service quality requirements. High-level data streams have higher priority than lower-level data streams when queuing and occupying resources. Differentiated services only contain a limited number of business levels, with a small amount of status information, simple implementation, and good scalability. Therefore, it is the IP network QoS solution recognized by the industry at present.
In the case where NGN services and Internet services share network equipment, the impact of Internet service traffic characteristics on network QoS performance should be fully considered. All along, the network planning and construction of the Internet basically refer to the empirical model of average statistics. While fully considering the business development needs, a certain margin is reserved. The capacity and bandwidth of the network equipment often exceed the actual needs, but in fact found The quality of the network is not stable. According to research in recent years, it has been found that the main services carried by the Internet such as WWW, FTP, etc. have self-similar characteristics within a relatively large time scale. The process of generating aggregate traffic is significantly bursty, rather than the traditional model. It is an expected smooth stacking process. The phenomenon concretely reflected in the IP metropolitan area network is that the average performance is better and the transient characteristics are poor. Therefore, the impact of Internet traffic on NGN services must be fully considered in network design. To solve this problem, DiffServ technology can be used to partially improve, or you can consider building a dedicated NGN network to avoid this problem. From an operational point of view, considering sufficient transmission resources, network bandwidth, low equipment cost, security and other factors, building a dedicated light-load network and increasing the network bandwidth as much as possible is also a very realistic solution.
On the basis of the above analysis, the QoS solution ideas of this paper are proposed, and the above technologies are mixed according to the actual situation. MPLSVPN technology is used at the backbone layer, DiffServ technology is used in the metropolitan area network and some private networks are built. At the nodes sharing equipment with the Internet, according to the entities carried by NGN, the priority fields of the NGNVPN LSP on the backbone network, the layer 2 VLAN of the NGN service, and the layer 3 IP address of the NGN service are all set to the highest level. The level field performs message classification, traffic shaping, traffic supervision and queue scheduling, so as to achieve high priority processing of NGN services, and minimize the impact of the sudden characteristics of the Internet on NGN services. In addition, the construction of special equipment is used to protect the NGN business in areas with a large amount of traffic, and to isolate the impact of the Internet on the NGN business.
For the design of service bandwidth, it needs to be calculated according to the VoIP traffic model. According to the characteristics of the NGN service we are currently developing and the actual network situation, we use the following bandwidth calculation empirical formula: NGN service bandwidth = [(number of users & TImes; single-channel bandwidth) & TImes; convergence ratio & TImes; (1 + 2.5% )] / 0.8, where speech channel bandwidth = coding rate + packet header overhead / packaging period; convergence ratio = speech channel convergence ratio × mute compression ratio (if VAD, take 60%, otherwise take 1); speech channel convergence ratio: 10,000 If users are above 0.1, 10,000 to 1,000 will be 0.25, and below 1,000 will be 0.5. The convergence ratio can be adjusted according to the development of local services and the actual situation of the network.
900KW-2000KW Brushless Generator
Electric Generators Direct,900Kw Brushless Generator,2000Kw Brushless Generator,900Kw Brushless Commercial Generator
Jiangsu Lingyu Generator CO.,LTD , https://www.lygenset.com