Voice Over MPLS switching router
Original Publication Date: 2002-Dec-25
Included in the Prior Art Database: 2002-Dec-25
Related PeopleOther Related People:
The invention describes a concept and a technical specification of a new architecture and product for transporting and switching of voice over MPLS (Multi Protocol Label Switching) network. Furthermore it describes how a traditional MPLS network can be enhanced with some new features that make is possible to transport and switch voice over MPLS in a scaleable way with guaranteed QoS (Quality of Service) and a low delay. The voice over MPLS switching router is a device by means of which a voice network can be constructed on top of an MPLS transport network. The MPLS packet based network is used for the transport of voice as well as data. For the transport voice or data label switched paths (LSP) are set up between different routers in the network. On top of the LSP’s voice samples can be carried according to the specification defined by the MPLS forum. Multiple voice channels can be mapped into a single MPLS label according to this standard. The purpose of the VoMPLS switching router is to be able to perform switching between the different channels that are available on different VoMPLS LSPs that go in and out of the router. The main objective of this product is to allow the construction of a very efficient next generation net-work that is able to transport voice as well as data. The voice is transported with a guaranteed QoS by means of this product. MPLS provides means for traffic engineering, path protection and QoS for the LSP’s that are set up between routers. By transporting voice directly over MPLS instead of transpor-ting voice over IP a lot of header overhead is avoided. With this solution it is not necessary to set up a new LSP every time a voice call is made. Instead of using one LSP end-to-end for transporting the voice samples, the voice samples will be sent over a set of different MPLS LSPs when they are transported from one IP domain to an other IP domain. At some points in the path the voice samples need to be decapsulated from the Voice Over MPLS packets and need to be rearranged in new LSPs. Within the network we need thus two kinds of equipment. The first type is a normal MPLS router. This router does not need to be aware of the fact that voice or data is stored in the payload of the MPLS packets. It just needs to look at the MPLS label in order to do the forwarding. Voice packets are just prioritized in these routers based on the EXP bits in the MPLS label or based on the priorities and bandwidth that were reserved during the setup of the LSP. The second type of device we need in the VoMPLS infrastructure is also an MPLS router but with ex-tended functionality. This router will forward the packets from the LSPs for which it does not perform a voice switching task in exactly the same way as a normal MPLS router. For those LSPs for which it wants to perform a voice switching function it will terminate the LSP, decapsulate the Voice over MPLS and rearrange the voice samples in new LSPs according to the direction that fits with their final destination. For this purpose a new kind of label switching table for voice needs to be added to the router which is explained further on in the text. Figure 1 shows a network architecture which could be used for establishing voice calls in an efficient way over an MPLS infrastructure. The different LSPs shown in this picture are part of the path that the voice samples will follow. They are comparable to trunks in a traditional voice network. The LSPs are fixed configured and dimensioned in bandwidth according to the long term traffic planning. These voi-ce LSPs are assigned a higher priority than data LSPs during the LSP setup with LDP of RSVP (Re-source Reservation Protocol). Packets arriving on these LSPs are assigned to low latency queues in al MPLS enabled routers in the network. In a normal MPLS router a voice over MPLS packet is not changed and just forwarded.