QoS technology of VoIP in MPLS-Based Integrated Model (1)

1 Introduction

VoIPVoice over IP is a technology that uses an IP network for voice communication. As IP technology is a connectionless technology, the original intention of an IP network is to provide a service called "Best Effort" and "Best Effort, this is suitable for data services that only require accuracy and have no strict requirements on latency. For real-time communication services such as voice and video, their Service Quality is of Service and QoS) it is hard to guarantee. The VoIP service is mainly attributed to the bearer network problem. The current network bandwidth limit is the main cause of high latency and congestion. In addition, when both voice and data applications are provided in a network, the service quality of voice applications must be taken into special consideration.
To ensure the QoS of the IP address, IETF first proposes a comprehensive service model Intserv) [1] Using RSVP to send signals. Before sending data, it establishes a path for the receiver and Reserves Resources, end-to-End QoS is achieved through mechanisms such as admission control, policy control, and classified scheduling control. To reserve resources for each stream on each node and create and remove paths, each node must support RSVP, it is necessary to maintain the "soft state" information of routes and resources, so that it has poor scalability and robustness. It is especially difficult to implement large-scale Wan on the existing network. This prompted LETF to develop a differentiated business model (Diffserv) [2], which resolves a business flow to a small volume of aggregated streams at the edge of the network ), it is identified by the DSCP Diffserv Code Point of the IP grouping header, which identifies the Service Code. It implements classification, marking, management, and other functions at the network edge node, the core nodes in the network are only forwarded to the group based on the PHBper-hop-behavior related to DSCP. This simplifies the structure of nodes in the network, which is much more scalable than the overall service. However, Diffserv still uses a hop-by-hop packet forwarding method, which does not provide end-to-end QoS support.
For VoIP, the Internet must have two basic attributes: QoS Assurance and optimal resource use. Optimal use of resources is a necessary step to avoid traffic congestion and service degradation. This work is done by the traffic engineering. Multi-Protocol Label Switching, MPLS) is widely regarded as an important traffic control tool for IP networks. This importance comes down to two main features: first, the use of short and fixed-length labels during packet transmission improves the presentation performance. Second, the capability of creating a circuit label switch path, LSP) does not need to be linked to [3] in the network. These MPLS features are available in both Intserv and Diffserv.
From this we can see that Intserv/RSVP, Diffserv, and MPLS are complementary technologies in pursuit of end-to-end QoS. Therefore, to ensure the QoS of VoIP, this integration model is used, Intserv is used in the edge network, and Diffserv Over MPLS is used in the core network. This article discusses the QoS technology for transmitting VoIP services on this integrated model.
2 MPLS
2.1 MPLS Introduction
MPLS is a Multi-Protocol Label conversion technology. It has the advantages of layer-2 packet forwarding and layer-3 route selection, the aim is to solve many problems of the group forwarding technology used in the current network environment. The essence of MPLS is that when an IP packet enters the MPLS network, it is assigned a short, fixed-length, locally meaningful label that can distinguish it from other information flows as an MPLS header to encapsulate this IP packet, all forwarding mechanisms in the MPLS network are based on this label, which tells the switching node on the group path how to process and forward data, and unencapsulates the MPLS header when leaving the MPLS network. The MPLS header contains a 20-bit label, a three-bit extended domain is initially defined as extended, and is now used as COS-service type domain), and a bit of label stack indication, there is also a bit of TTLtime-to-live) domain.
MPLS has several core technologies and components: Traffic Engineering, constraint-based routing, Label Switch Router, LSR), Label, Label exchange and Label distribution, LSR is a vswitch or vro that distributes tags and can be grouped based on tags. In an MPLS network, the switching path can be a point-to-point, multi-to-one, one-to-many, and multi-to-many path.
2.2 LSP and traffic engineering on MPLS
All groups access the MPLS network through the ingress LSR and exit the MPLS network through the egress LSR. This mechanism creates LSP, which refers to the specific FEC, A set of LSR tag sequences that must be passed by a tag group before it reaches the egress LSR. This LSP is unidirectional, that is, different LSP will be used to return data streams in a specific FEC.
The establishment of LSP can be a control driver, that is, triggered by the control traffic), or a data driver, that is, triggered by the emergence of a special stream ). The ing between the IP package and the LSP must take place at the LSR entry by specifying a FEC for a tag. The LSR Entry uses a feing from FEC to NHLFENext Hop Label Forwarding Entry), which is used when the forwarded packet has no labels and will be marked before Forwarding.
To establish LSP, LSR uses signaling information to coordinate and distribute tags. These signaling information can be carried either using a new Protocol called LDPLabel Distribution Protocol or using an extended RSVP [4. The two protocols provide similar functions on creating LSP and supporting traffic engineering constrained routes. When transmitting VoIP streams in an MPLS network, extended RSVP is generally used to distribute tag binding information.
The traffic engineering can move the data stream from the shortest short circuit calculated by the routing protocol, so as to arrange the data stream to pass through the network, avoiding the obstruction caused by the uneven use of the network. Therefore, it has many benefits to implement traffic engineering on an IP network, mainly in two aspects: Traffic-based and resource-based. The former is the key traffic execution features, such as latency, packet loss, and throughput efficiency. The latter is the most effective way to use available network resources to avoid congestion and low utilization. The direct advantage of using traffic engineering technology is to avoid congestion points when forwarding traffic. In case of failure, you can quickly reselect the route to effectively use available bandwidth and QoS.