7.7 Session and Transport Issues


In IP wireless networks a clear separation between session and transport is introduced. Traditionally, a session is a relation between two nodes that is established between the two nodes by means of some protocol and has a limited lifetime.

When the concept is applied to IP wireless networks, the term session refers in particular to application sessions (i.e., temporary associations created between two or more IP nodes [e.g., two IP terminals and one server, one IP terminal and one server] in the framework of a given application). Examples of sessions are the relation between an IP node and an e-mail server to support the retrieval of e- mails by the user (e-mail session), and the relation between two SIP nodes and one or more SIP servers (SIP session). In general, session refers to the set of services the IP node can access.

A session foresees transfer of IP packets (signaling or user plane) between the nodes in the sessions. Therefore, each session is supported by transport. What is traditionally meant by the term transport in reference to a session is a specific flow of IP packets (also referred to as IP bearer) or set of flows that are exchanged between two IP nodes. The transport carries the information to be exchanged between two IP nodes as result of application sessions. Transport can also be unidirectional.

In particular, the following distinctions can be made for transport:

  • End-to-end transport: Represents IP flows exchanged between two IP nodes, namely the source and the recipient of IP packets in a communication.

  • Wireless access transport: A wireless access session is a given bearer that supports an end-to-end session. In case of cellular networks, the wireless access session can be a set of radio bearers, or in the case of 3G UMTS packet data networks, a set of PDP contexts and the associated radio bearers .

Figure 7-5 depicts the differences between session and transport.

Figure 7-5. Application sessions and transport.

graphics/07fig05.gif

In IP wireless networks, session and transport can be considered as separated, as depicted in Figure 7-5, in contrast to the close relation that exists between signaling/control and user plane information in traditional cellular networks. In traditional networks, user plane and control signaling are processed by the same network entities (e.g., BTS, MSC in GSM). The adoption of IP in data networks introduces the ability to carry signaling information and user plane data separately in the network, therefore delivering the information to separate network elements. An example of this is the ability to run SIP services over IP wireless networks (e.g., General Packet Radio Service [GPRS]).

Due to the separation between session and transport, a new set of issues arises. In particular, two issues are relevant and are closely interrelated:

  • The control of session and transport

  • The ownership of session and transport

The following section describes these issues.

7.7.1 Control and Ownership of Session and Transport

Figure 7-6 describes the relation between the control of session and transport by depicting two examples. In one case, the IP node is involved in an application session where both the control of the session and the transport of IP packets are provided by the same network operator. This scenario is very close to the way services are provided in cellular networks. In the other example, the IP node is accessing services provided by a third party (e.g., Internet service provider) through the transport provided by the network operator. In this case, session and transport are not controlled by the same party.

Figure 7-6. Relation between application and transport sessions.

graphics/07fig06.gif

The separation between session and transport is perceived by several service providers as a threat to their business. In current cellular networks, users access services through the same operator that is providing the connectivity to the network, therefore guaranteeing operators complete control of service provisioning. With the advent of IP and Internet applications and the separation between session and transport, traditional operators are concerned that they will lose control of service provisioning (i.e., lose ownership of sessions and be relegated to the role of wireless ISPs).

In this framework, solutions need to be designed to allow two service-provisioning models:

  • Wireless IP operators provide complete services, therefore maintaining control of sessions. To enable this model, operators must be able to control how transport, and in particular QoS, is provided to IP nodes. In fact, IP nodes could request IP transport with QoS to the network operator and access services of third parties.

  • Wireless IP operators provide complete services and simple IP connectivity to IP nodes (i.e., transport with possibly quality of service).

Only the availability of solutions enabling both models would facilitate a successful deployment of IP wireless networks.



IP in Wireless Networks
IP in Wireless Networks
ISBN: 0130666483
EAN: 2147483647
Year: 2003
Pages: 164

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net