4.5 N ETWORK T ECHNOLOGIES
4.5.2 Service Networks
Service networks provide the final service to the user. The service that is offered by the transport layer is used to interconnect the nodes that form the service network.
In 1978, the ISO (International Standards Organization) started the works to define an open communication architecture. After ten years of works, the OSI (Open System Interconnection) model was released. The OSI model defines a generic architecture for data communication networks that due to its global and wide perspective is normally used to explain the operation of communication networks.
Two types of systems have been defined in the OSI reference model: End-Systems and Intermediate Systems.
An End-System is a terminal equipment that delivers the final service to the user. An Intermediate System is a network device, which does not directly support users but forwards received data towards the final destination. Intermediate systems do not need to understand the information being sent between the users, but need to understand and possibly modify the information added by the network to provide the communication.
End Systems may be directly connected, but more normally rely on the service provided by one or more Intermediate Systems. Examples of intermediate systems are routers or network switches.
The communications process between End systems and Intermediate systems is usually defined in terms of the seven layers OSI reference model. In this reference model, shown on Figure 4.5-3, intermediate systems handle only protocol information at and below the network layer, whereas end systems use protocols at all the layers of the reference model.
Application Programs
Physical Layer Physical Layer
Data Link Layer Data Link Layer
Network Layer Network Layer
Transport Layer Transport Layer
Session Layer Session Layer
Presentation Layer Presentation Layer
Application Layer Application Layer
1 1 2 2 3 3 4 4 5 6 1 1 7 7 6 5 2 2 Medium Medium Application Programs 3 3
Every layer has a well-defined functionality and provides a service to the upper layer in the model. The functions carried out by every one of the OSI model layers are:
1. Physical layer. The responsibility of the physical layer is to transmit unstructured bits of information across a link. It deals with the physical aspects such as the shape of connector, pin assignment, etc.
2. Data link layer. The responsibility of the data link layer is to transmit the information across a link. It deals with error detection and correction, information alignment and addressing when several system are reachable as in LANs or multipoint links.
3. Network Layer. The responsibility of the network layer is to enable the communication between any pair of end system in the network. The network layer deals with the route calculation function, congestion control, etc.
4. Transport layer. The responsibility of the transport layer is to establish a reliable communication stream between a pair of End systems. It deals with the detection and correction of the errors introduced by the network layer, such as packet lost or duplicated, reordering of out-of-order information, etc.
5. Session layer. The responsibility of the session layer is to co-ordinate the way data are transferred throughout the communication provided by the transport layer.
6. Presentation layer. The responsibility of the presentation layer is to adapt, when necessary, the different internal data representation format used by the End system that are transferring information.
7. Application layer. The responsibility of the application layer is to deliver the communication service to the application that is using the service provided by the network.
Service networks can be implemented by using different technologies. Every technology in general could be more suitable to offer some services. Since the design of modern network is focused towards service integration, only those technologies that can be able to offer service integration are being considered for future designs. Nevertheless, we are going to mention not only future trends but also existing technologies owing to their capabilities to support the related applications.
4.5.2.1 Circuit Switched Networks (POTS, ISDN)
Circuit switched networks are connection oriented networks. The establishment of a connection requires a call set-up that chooses a path in the network in which the necessary resources to support the connection are reserved.
Resources are allocated to a connection whilst this connection is maintained, even though they were not used. Only when the connection is released will the resources be liberated.
Circuit switched networks can be based on analogue transport technology, on digital transport, whether PDH or SDH, or on a mixed configuration. Due to the fact that a circuit of constant bit- rate, usually 64 kbit/s, is used to support the connection, a deterministic delay performance is achieved in the final service offered. On the other hand, since every connection established in the network is based on the use of 64 kbit/s channels, when a connection is used for the transmission of information with a lower bit-rate poor resource efficiency is obtained, unless sub-multiplexing techniques are applied.
4.5.2.2 Packet Switched Networks (X.25, Frame Relay)
Packet switched networks are connection-oriented networks. The establishment of a connection requires a call set-up that chooses a path in the network in which the necessary resources to support the connection are reserved. The difference with Circuit Switched Networks is that network resources are shared by the users, that is to say, resources are only used when needed. Thanks to this, a 64 kbit/s channel can be shared by several connections of lower bit- rates.
This mechanism allows resource optimisation to be achieved at network level but, on the other hand, a non-deterministic delay is obtained for every virtual connection due to the effect of statistical multiplexing used in the network.
This type of networks offers packet-oriented services. It is not possible to obtain constant bit- rate services due to the intrinsic non-deterministic delay of its transmission mode. They are mainly used to interconnect computers as they offer data communication services.
X.25 and Frame-Relay networks are examples of this type of networks. They provide data packet communication service primary used for LAN or Mainframe interconnections and data network implementation.
4.5.2.3 Cell Switched Networks (ATM)
Asynchronous Transfer Mode is a very efficient switching technology that has been adopted by the ITU-T as the base for the Broadband ISDN (B-ISDN) network.
B-ISDN is a connection-oriented network. Thanks to the use of ATM, any type of service such as packet-oriented, circuit-oriented, constant bit-rate, variable bit-rate or even connectionless can be integrated on the same network.
In ATM networks, the information is carried in cells. The cells follow the pre-established path in the network. Cells are generated depending on the amount of information the user wants to transmit. Resource optimisation is achieved as cells are generated only when some information has to be transmitted, so that the network capacity is shared by the users, the total amount of bandwidth required being lower due to the statistical multiplexing gain.
One of the new concepts introduced by ATM is the flexible bandwidth and QoS service allocation, (see A3.2). It is possible for every user to set requirements on bandwidth, total End- to-End delay and delay variation. Thanks to this, connections with fixed bandwidth and bounded delay and delay variation can be defined in an ATM network. This possibility is used to offer the Circuit Emulation Service (CES), [14]. The performance of a circuit emulated by an ATM network is comparable to that experienced with current TDM technology.
CES offers structured DS1/E1 Nx64 kbit/s (Fractional DS1/E1, where a selected subset of the 32 channels from the entire frame are transmitted, i.e. N = 1…32) services as well as unstructured DS1/E1 (2'048 kbit/s gross data rate, transparent bit-by-bit transmission)
IWF
IWF
DTE
DTE
ATM CBR
Figure 4.5-4: Reference model of the Circuit Emulation Service (CES)
Figure 4.5-4 shows the reference model of the CES. We can distinguish three main components: The Data Terminal Equipment (DTE), the Internet-Working Function (IWF) and the ATM network. The DTEs are the actual users of the service, a protection relay or a teleprotection equipment providing 64 kbit/s or N times 64 kbit/s for instance. The IWF provides the conversion of the bit-stream generated by the users into cells and the reconstruction of the original bit-stream at the reception side including timing recovery and jitter removal. This function is usually embedded in the ATM access device. Finally, the ATM network should provide a Constant Bit-Rate (CBR) virtual channel that should have been dimensioned with the fixed bandwidth required to carry the bit-stream provided by the users.
When Nx64 kbit/s working mode is selected, cross-connect (DXC, Digital Cross-Connect) functionality can also be provided, thereby being possible to deliver every single 64 kbit/s channel to different locations in a similar way as PDH cross-connect devices do.
Since the CES has to offer quality performance similar to a PDH/SDH connection, it has to fulfil the requirement of the related standards. Therefore, a CES has to comply with ANSI T1.403 and ITU-T G.824 for jitter and wander performance of digital networks that are based on the 1544kbit/s hierarchy, and ITU-T G.823 for networks that are based on the 2048 kbit/s digital hierarchy, see also Table 4.4-1. Other facilities related with the data format and structure such as framing, alarm transmission, loops, etc, should comply with the relevant standards already applied to PDH/SDH connections.
The Bit Error Rate (BER) of the emulated channel should comply with the ITU-T G.826 recommendation for E1 (2048 kbit/s) and the ANSI T1.510-1994 for DS1 (1544 kbit/s), or ITU-T G.821 for lower bit rates, e.g. 64 kbit/s.
CES could find a direct application to connect existing protection relays or digital teleprotection equipment to ATM networks without the need of a specific implementation or external adaptation devices. Nevertheless, no practical experience of using this approach has been reported until the moment of writing this document.
4.5.2.4 Datagram Networks (IP)
The traditional network concept we have discussed so far is based on the circuit-switched approach. Each connection is associated to a circuit that has resources allocated for its exclusive use along a path. There is no uncertainty about the bandwidth or delay along this path so the Quality of Service in terms of bandwidth and delay can be guaranteed.
packets are delivered to the Network without any resource allocation, and the network exerts its “best effort” to serve the packets.
Two different working modes can be distinguished in data networks, “Hop-by-Hop” control, used in Virtual Circuit Networks and “End-to-End” control used in Datagram Network.
In the first approach, a connection is set up in the network, so that every intermediate system involved in this connection change its internal state. Every node takes care of every packet and guarantees its transmission towards its destination. This scheme suffers from a side effect known as “fate-sharing”: The End-to-End connection depends on the state of all the intermediate systems involved in this connection. If any of those systems fails, the connection will be lost.
In the datagram approach, the End-to-End connection does not depend on the state of any one of the intermediate systems of the network. If one of the intermediate systems fails, the information will be routed using another path so that the final users will not be aware of this change. This scheme increases the overall availability of the network with this effect being more important for bigger networks.
In the End-to-End scheme, the responsibilities are shared between the network and its users. The Network is responsible for the routing whereas the users are responsible for the control of the communication. Thanks to this approach, the Datagram networks present an unmatchable resilience level as well as the best resource optimisation. These characteristics make them suitable for mission-critical applications such as most of the applications that can be found in the Power Utility Control Network environment.
Datagram networks using IP (Internet Protocol) cannot assure the QoS as the network presents a non-deterministic transmission delay. Then it cannot be applied to delay sensitive applications such as teleprotection, unless some specific Quality of Service mechanisms were added in order to guarantee bandwidth and/or delay. Refer to chapter A3.3 in ANNEX A3.
The great flexibility of this type of networks makes them suitable for service integration. Although they cannot intrinsically offer a constant bit-rate service, thanks to a new application protocol that has been defined (Real-time protocol) it is possible to eliminate the delay variation at the application level. However this comes at the expense of an additional delay that may not be acceptable for protection.