Unlike SIP, TIPHON is developing the ability for control of the routeing of the media channel as well as the signalling channel.
3.3.6 Proprietary protocols
Many ITSPs are using proprietary software for PC-Phone and PC-PC applications. The use of proprietary software is likely to continue for some time until standardised interoperable software becomes available at reasonable prices.
3.3.7 Comparison of protocol stacks
Figure 18 compares typical protocol stacks for signalling and call control.
Figure 18: Typical call control protocol stacks 
Figure 19 compares typical protocol stacks for media packets.
Figure 19: Typical protocol stacks for media packets 
In both the above figures the full range of options for the use or absence of ATM and SDH are not shown. They are explained in section 3.4.3.
3.4.1 Introduction
The main equipment suppliers such as Cisco, Lucent, and Nortel have invested many hundreds of millions of dollars in developing multi-service architectures and components for managed IP networks. These networks are often called next generation networks. They are seen by their suppliers as long term replacements for existing circuit switched networks and specialist data networks.
Figure 20 shows the basic architecture of a next generation network. There are three parts, which are sometimes thought of as layers:
-
The access system, which is designed to work with a variety of access transmission systems
-
The transport layer described in section 3.4.3
-
The services and control layer where the key component is the softswitch described in section 3.4.2
Figure 20: Next generation network  3.4.2 Softswitches
A softswitch is a set of software that provides:
-
service provision
-
call management (call server)
-
subscriber management
-
call record generation.
With the SIP protocol, it implements the proxy functions, with H.323 it implements the gatekeeper functions. Some designs such as Ericsson’s Engine run BICC, which is derived from the ISDN signalling system ISUP and adapted for use over ATM. Whereas call control was implemented in circuit switched networks using very expensive special purpose hardware, softswitches are implemented on general purpose computers such as Sun Workstations.
Whilst the softswitch provides the call control, calls may be routed to an announcement server, which is a voice response system that handles tones and any in-band functions.
3.4.3 Transport structures (ATM, IP and MPLS)
Network designers have to design networks to support one or more types of application with differing sensitivities to delay and with different variations in traffic loading. At the physical level they have a basic topology of physical cables or radio links that interconnects the nodes of the network. They then have up to three methods of sharing and interconnecting the physical capacity. The objective of the design is to maximise the traffic handling capacity within the quality constraints of the services being supported whilst minimising cost.
-
At the physical level (layer 1), they use the structures of the Synchronous Digital Hierarchy (SDH) to subdivide the capacity on the cable fibres and add some basic fault monitoring capability. At nodes they use cross connects to interconnect different units of capacity on different cables. This results in a different and normally more complex topology of capacity.
-
At the link layer (layer 2), they use a packet type system to create an even more finely grained system of links to time share capacity on the topology created by SDH. This is either the ATM system of short fixed length cells, or the frame relay system of longer and variable length frames. The ATM links connect ATM routers that switch ATM cells in accordance with a link identifier called a virtual path – virtual channel combination.
-
At the network layer (layer 3), traffic between end users is sent in IP packets that are routed by IP routers at layer 3 nodes according to the IP address of the destination.
IP packets can run on ATM that in turn runs on SDH or run directly on SDH, ie layer 2 can be null.
IP networks always have IP routers at their edges or access points to route incoming and outgoing packets. Inside the network, however, they can choose either to place other IP routers at the nodes or to use ATM routers to create a mesh of virtual connections between the IP routers at the edge. Although the use of ATM adds additional overheads in terms of the ATM headers, it has several advantages:
-
ATM routers introduce less delay than IP routers because they are much faster as they can be built with more functions in hardware, whereas IP routers depend more on software
-
ATM routers offer more flexibility in running multiple queues
-
ATM can distinguish between 5 different quality classes, whereas mechanisms for distinguishing quality in IP are not yet fully developed.
These advantages are especially valuable for conversational voice because it is highly sensitive to delay. Therefore many of the telcos are using ATM inside their networks because they see telephony as their main source of revenue and wish to maintain their competitive edge in voice quality (low delay).
There is a further recent development called Multi-Protocol Labelling System (MPLS)14. With MPLS, IP routers at the edges of a network attach locally defined labels to IP packets. These labels define a Forwarding Equivalence Class (FEC) and they are used by internal IP routers for:
-
Routeing. Label edge routers can use labels to define a whole route through a complex network of ATM nodes.
-
Distinguishing different classes of traffic for quality differentiation.
The use of labels reduces the processing load on the router because the label can be analysed more easily and quickly than the IP address. The three main advantage of adding label switching to an IP core network that does not use ATM are:
-
The reduction in delay
-
The addition of source route control
-
The introduction of different quality classes
Where the internal routers are ATM15 rather than IP, the label is carried in the ATM header instead of the virtual path – virtual channel identifier. The three main advantage of adding label switching to ATM are that:
-
MPLS has a well defined protocol for establishing labels for routes within a network
-
MPLS labels can be used with IP, ATM and frame relay and so can work across a network of mixed technologies
-
MPLS is better equipped for rapid reconfiguration for congestion control
For IP packets that relate to signalling and data, there are four options inside a network:
-
IP alone
-
IP with MPLS
-
IP on ATM
-
IP with MPLS on ATM
For media, there is the additional option not to use IP at all but to run the voice directly over ATM using the ATM Adaptation Layer AAL1 or AAL2. This is the lowest delay solution. Since the media stream in telephony is much more sensitive to delay than the signalling, the option of running the signalling as IP on ATM with the media going directly on ATM is quite attractive.
The design and management of IP/ATM networks is the subject of much development work at present.
SDH is also not essential and some companies are developing “IP over ” routers that put IP packets directly onto optical modulators.
Share with your friends: |
