ATM CONTROL PLANE

 

 

BACK TO	ATM TABLE OF CONTENTS
	CONTROL PLANE

 

The Control Plane is responsible, on user's behalf, for the control of:

• Switched Virtual Connections SVCs, either VPC or VCC
• Permanent Virtual Connections PVCs, VPC or VCC.

• Point to Point (symmetrical or asymmetrical)
• Point to Multipoint
• Multipoint-to-Multipoint

To support multimedia concurrent applications, an ATM call can control several simultaneous connections with their own QOS; they can be routed through the same physical path, with the Common Routing Group facility.

 The architecture of the control plane is described below.

SAAL: Signaling AAL

The user plane connections are setup and released by the control plane, by a signaling exchange between the ATM end systems and the ATM intermediate systems (switches). ATM is a Connection Oriented protocol, whereby user data can be exchanged only after a connection has been established. The transmission characteristics (QOS, throughput, latency) are fixed during the whole connection. To change them, the connection needs to be released and reestablished. Moreover, the signaling information and the user data do not share the same channel path; ATM uses an outband signaling scheme.

Connection control requires that each network entity (including the user terminal) be identified by a unique address.

ATM ADDRESSING

There are two standardized addressing schemes:

• E164 already specified by the ITU for ISDN and to be used in public networks
• NSAP (Network Service Access Point) defined by the ATM Forum for private network

NSAP is more complicated to take into account different numbering schemes in use by private organizations.

NSAP	IDP		DSP
	AFI	IDI	HO-DSP					ESI	SEL
	39	DCC	DFI	AA	HO-DSP*			ESI	SEL
	47	ICD	DFI	AA	HO-DSP*			ESI	SEL
	1 B	2B	1B	3B	6 B			6 B	1 B
	45	E164				RD	AREA	ESI	SEL
	1 B	8 B						6 B	1 B
IDP: Initial Domain Part - DSP: Domain Specific Part - AFI: Address Format Identifier - IDI: Initial Domain Identifier - HO: High Order - ESI: End System Identifier - SEL: Selector - DCC: Data Country Code - DFI: Domain Format Identifier - AA: Address Authority - ICD: International Code Designator - RD: Routing Domain

It does rationalize all these schemes by defining a standard format. The AFI field identifies different addressing schemes, among those:

• DCC (AFI=39) with each country having its own IDI and the addressing plan administered by the ISO National Member Body.
• ICD (AFI=47) identifying different international organizations, administered by the British Standards Institute.
• E164 compatible

In ATM it is also possible to identify not only single users but also groups of users. ATM signaling does support multicasting and anycasting.

• Multicast consists of sending the same information to a group of end-users. It's based on "Group addressing" yet in the standardization process
• Anycasting , also based on Group Addressing, allows an end user to be connected to any group member (emergency number for instance)

ATM SIGNALING & ROUTING

As said before, connections between end users are set up and released by the signaling linking the user to the network and, within the network, the different switches. In fact rather sophisticated connection services can be provided as underlined hereafter.

 

Basic Services	Supplementary Services
. Point-to-Point connection set up &release  . VPI/VCI selection & assignment  . Quality Of Service class request  . Traffic parameters request  . Subaddress support  . Identification of calling party  . Transit Network Selection  . Basic error handling  . User-to-user signaling	. Point-to-Multipoint  . Symmetric operation  . Multipoint-to-Point or Multipoint  . Multiple connections setup  . Call Transfer  . Call Forwarding  . Call Offer  . Call "Do not Disturb"  . Multiple Subscriber Number  Etc.

Of course not all of this is available to day. ITU and the ATM forum continue to work to make these services standardized, available to end users.
The service enhancements are released through so called Capability Sets: CS2 was available in 1995, CS3 is scheduled for 97.

To provide those connection services, the ATM end system and network nodes use a signaling protocol, specified by the ITU for the public ATM networks and the ATM Forum for private ones. There are two different versions: UNI (User-to-Network) and NNI (Network-to-Network)

SIGNALING PROTOCOL

• The "signaling message and sequence" layer generating and processing the signaling messages (exchanged between the ATM communication entities) to implement the connection services requested by the end-user. UNI and NNI use different layer protocols, with UNI relying on the Q2931 protocol (derived from the ISDN Q931), NNI on SS7, the internal protocol of the new digital telephone networks (and the intelligent networks).

The following is focused on UNI.
The UNI Q2931 protocol defines the messages to be exchanged: their content and their sequence. For instance, the connection setup and release message sequences are shown below.

The signaling message is composed of so called Information Elements (IE):

• "AAL parameters", such as AAL type, SSCS type and CPCS-SDU or MTU (Message Transfer Unit) values
• "Broadband Low Level Information (B-LLI)", equivalent to ISDN LLI, defining which Layer 2 & 3 protocols will be used in the SVC
• "ATM Layer" parameters, such as the traffic service class, with the QOS & traffic descriptor values, the ATM addressing information (called and calling parties with their local ESI address)

Other Information Elements are being standardized: Broadband High Layer Information (B-HLI) to ensure interoperability of terminals, Network Layer Information (B-LNI) for network interworking.

 

• The SAAL (Signaling AAL) layer is responsible for the reliable end-to-end transport of the signaling messages. It relies on AAL5 and consists of two sublayers:
• The Service-Specific Coordination Function (SSCF), specified in ITU Q2130, mapping Q2931 signaling onto SSCOP, the next SAAL sublayer
• The Service-Specific Connection Oriented Protocol (SSCOP), using the AAL5 services, provides for reliable transport with error recovery and flow control, for establishing, monitoring and releasing the signaling session.
• The ATM and Physical Layers are the same as for the User Plane but dedicated channels are used for signaling, usually the VPI=0/VCI=5 channel. Indeed, a fixed and well-defined channel has to be set between two consecutive ATM signaling entities to exchange signaling information.

• Soft PVC allows PVC-capable end-systems to be easily supported across a SVC-capable ATM network. It avoids tedious manual configuration of the PVC within the network and may provide also for rerouting in case of link or switch failure.
• SVC tunneling is useful across an only PVC-capable ATM network. Two ATM SVC end-networks will be interconnected through a tunnel (PVC) on which the signaling channels will be conveyed. Two different VPs will be required to connect two ATM end-switches: one for the user data, one for signaling.

ROUTING

The purpose of routing is to find a "route" between the calling and the called parties that fulfills the user requested connection traffic contract. A route is a succession of VC/VP links (identified by the VPI/VCI cell header field) making up the VCC/VPC connection. Each switch holds a routing table defining for each connection the relationship between the input VCI/VPI link and the VCI/VPI output link
 

Different routing algorithms can be used, either static or dynamic.

• With static routing, the switches are interconnected through fixed internal routes (fixed routing table). Thus, in reference to the next figure, to go from User U1 to User U2, the connection will be established via Switches A and B. If unfortunately the A-B link breaks, U1 cannot communicate with U2 although there is an alternate path A-C-B.

 

A refinement to static routing is alternate routing where an alternate route backs up the fixed route. In our example, Switch A will have two possible routes to reach B either directly (via A-B) or via Switch C. When Switch A will receive a call set up message (U1to U2), it will check if the A-B link is operational. If yes, it will forward the Setup Message to Switch B and establish a VC link with B. If no, the A-C link will be used and a VC link established with Switch C. C will further process the Setup Message and establish a VC link to Switch B.

• With dynamic routing, for each user requested connection, the best route fitting the requested traffic contract will be found out. In our previous example, Switch A will select the A-B path or the "A-C & C-B" path on the basis of the traffic loads on those paths. Therefore U1 could communicate with U2 via A-B at a given time and later, at the next connection, via A-C-B depending on the network load.

To achieve it, the switches A, B and C shall exchange topology and network load information. Switch A learns that, to reach U2, two paths are possible (A-B and A-C-B) and knows at each instant the traffic conditions of these two paths. It will then select the proper one when U1 wants to communicate with U2.

The ATM Forum has standardized a dynamic routing protocol, the PNNI (Private Network-to-Network Interface) to be used in ATM end systems and ATM switches. It does carry not only reachability information but also trunk and switches traffic capabilities to ensure "QOS & traffic descriptor" specified connections. It is based on an hierarchical network structuring, whereby the network is divided into domains, themselves into "peer groups". Each switch in a group has detailed information on his peers but not on the outside except one of them so called the Peer Group Leader that represents (with summarized reachability and traffic load information) the group in a higher order peer group (its parent). It forwards to his peers the summarized information he gets from is parent group.
This structuring is comparable to the that of the telephone network with different switches hierarchies: local switch, area switch, transit switch (national and international). It relieves any individual network switch to store and process all the reachability and traffic conditions of the whole network, which would be an enormous amount if the network would consist of several hundreds of switches.

After network routing information collected, a given switch must, at the reception of a setup message, decides which route to use. This is done by the so-called Connection (or Call) Admission Control (CAC) algorithm. PNNI standard has replaced it by the Generic CAC

 

BACK TO	ATM TABLE OF CONTENTS
	CONTROL PLANE