ATM Signaling Project

Introduction
Protocols
Implementations
People
Presentations
Related Links

Introduction

Signaling is a part of the network control system. A signaling system is responsible for connection setup and tear-down. An ATM signaling system has to convey diversity of ATM service parameters such as QOS, traffic information through the ATM network. At each network node, resources has to be checked to decide if the connection can be accepted. Maintaining routing information is another job of the signaling system. These multiple tasks make the ATM signaling system a complicated system. ATM signaling consists of User-Network Interface signaling (UNI) and Network-Network Interface signaling (NNI) as shown in the figure. An UNI is an interface between an end system and a network. A NNI is the interface between network nodes (or network clouds).

ATM Signaling is a highly standardized area. ATM Forum and ITU-T are the two most important standard bodies. ATM Forum specifies UNI3.1, UNI4.0 for UNI signaling, and PNNI for network signaling. So far ITU-T only specifies UNI signaling. No NNI signaling protocol is defined by ITU-T.

The signaling project at Washington University is a part of the Washington University's Gigabit Network Technology Distribution Program. The goals of the project are:

1) Implement standardized UNI signaling and NNI signaling. The project will implement ITU-T Q2931, Q2971, ATM Forum UNI4.0 and PNNI.

2) Easy to use. The project will support simple API for users.

3) Easy to modify for research purposes. We will open source code to research community to encourage more signaling protocol research.

4) Portability. The software system should run on major UNIX systems.

5) Scalability. The product of the project is not a research toy. It will control real ATM network. Due to hierarchical scalability of PNNI (does someone says that seven layers of hierarchy can cover the communication system of the whole world?), the scalability of the product is predictable.

We build our signaling software based on Naval Research Laboratory’s (NRL) ATM signaling projects. NRL’s signaling software implements signaling protocols as a simulation system. We extend the system to control real switches. Current plan is to control the Gigabit ATM switches and new Burst Switches developed at Washington University. The result signaling software will be a part of the ATM campus testbed.

Signaling Protocols

ATM Forum:

UNI 3.0. The earliest UNI signaling protocol that support only point-to-point connections.
UNI 3.1. Significant modification of UNI 3.0. Support point-to-multipoint connections.
UNI 4.0 Support Leaf Initiated Join multipoint connections
Private Network-Network Interface (PNNI) for network node signaling

ITU-T

Q931. Signaling protocol for narrow-band ISDN
Q2931. Point-to-point signaling protocol for both ISDN and B-ISDN
Q2971. Addition to Q2931 to support point-to-multipoint connections

Implementations

The implementation of the signaling system is based on NRL’s PROUST project and the Gigabit Switch Controller developed at ARL. NRL developed a set of object oriented software structures suitable for application level protocol development which they called FrameWork. PROUST is a PNNI simulation based on the NRL Framework. Next figure shows the internal structure of the signaling software.

Each port of a switch has three channels to the signaling controller: 1) ATM signaling messages go through the signaling channel; 2) PNNI Hello and Topology Update messages go through the PNNI channel; 3) ILMI messages go through the ILMI channel. Upwards messages demultiplexed at the Lower Port Mux. Each port has a protocol stack to process the messages coming from that port. In the middle of the protocol stack is a Call State finite state machine. Each call has its own FSM. A PNNI setup message goes through the Port Protocol stack and reaches the control processor which all the routing and admission decisions are made. The detail of the software will appear in a technical report. If you are interested to the software implementation, you can talk to dakang@arl.wustl.edu for detail.

People

Presentations