Purpose of Networks and Network Design

As discussed above, there are several types of networks seen from a functional point of view. A purpose of this par­titioning or layering into several functional networks is to help us to build network architecture with well-defined functional areas and interfaces between these that will simplify extensions and modifications to be done. Another purpose is to support the building of cost-effective net­works, both from a purchase and from a life-cycle point of view, with low operation and maintenance costs.

The end users of a telephone network have a more holis­tic view: They want to be able to contact each other so that they can talk when they are physically far away. Hence, the main purpose of a telephone network is to establish com­munication between people. In a mobile multimedia net­work this interest extends to having mobile access wher­ever the user is located also to media servers including ordinary databases and file systems as well as audio and video stream content.

End users of telephony not only want to talk, they also want to talk inexpensively and get a good quality of service. This requires efficient use of resources and a good network design. The costs to operate a network depend much on how the subscribers are distributed in the geographical area that the network is aimed to cover. Charging policy, traffic intensity, and traffic patterns such as length and locality of calls are other factors that all will influence the design of the network. Most networks are built in a more or less hi­erarchical structure to collect, concentrate, transmit, and connect the traffic such that higher-capacity transmission and switching equipment can be utilized. This reduces the costs for transmission and switching since many users can then use the equipment more efficiently and with less risk for blocking due to statistical multiplexing effects. The de­gree of concentration that can be utilized is influenced by the traffic intensity and the desired service quality level.

Another important cost factor is that the total physical length of the transmission path can be reduced in this way, and this is important since the cost for digging ditches and use of pipes is a large contributor to the overall network costs. Microwave links are used to reduce cost where it is too expensive to dig ditches. For high traffic routes, direct routes are often used, especially if the distances are short, such as in metropolitan areas. Alternative routes are also used to improve reliability, but these increase the cost and for that reason are usually not used in the access network.

Using the structural perspective of networks, one can for a specific area identify a distribution of subscribers that must be interconnected in a reasonably efficient way—for example, in a star — or ring-structured physical access net­work. One can then allocate a number of network nodes of different basic types that are needed, both to deal with ex­pected maximum traffic loads and to interconnect these in a way that keeps transmission costs close to a minimum. Typical nodes in fixed and mobile telecommunication net­works are illustrated in Fig. 2.

We can now go back to the external functional network perspective and hence to the different functions and ex­amine how these are distributed in the physical network. This distribution can be guided by different principles. One principle that has to do with costs is that simple and cheap functions that are used often and for long periods of time should be located close to the subscribers. Complex expen­sive functions used seldom and for short periods of time should be placed more centrally in the network and thus be shared by many users. However, with today’s technol­ogy and using mainstream components, complex must not mean expensive and hence a complex function if imple­mented as an integrated circuit can often be placed close to the subscribers, for example voice coders. Furthermore, the effects of distribution on reliability, signaling, and mainte­nance must also be considered when deciding on function distribution over network nodes.

From the above discussion one can see that there is no absolute definition of how networks shall be implemented or what a network node shall contain. Rather, there are several possible configurations that can fulfill the require­ments, because not only does the distribution of subscribers as well as their quality of service (QoS) requirements and traffic patterns differ significantly but also implementa­tion costs for different types of solutions differ and change over time.

Figure 2. Typical nodes in a telecom network.

Going back to and analyzing what an exchange is, one can see that many of the functions carried out in the dif­ferent logical networks have traditionally been placed in an exchange node and can still be located in such a node. On the other hand, these functions and hence the exchange
can also be more or less distributed and placed in more spe­cialized networks and nodes. With well-defined functional areas and interfaces between these it is possible to config­ure networks and network nodes in many ways. However, for a node or cluster of nodes to be regarded as an exchange it must have some basic call access, control, switching and connection handling capability.

Routing and Switching Techniques

The task of finding a path from source to destination is called routing. The task to follow such a route from source to destination guided by end-to-end address information is called route selection or forwarding. Each node that partic­ipates in this task does not know the whole path but must be able to analyze the destination address and find out in which direction the path should go and also find such a path with a trunk that is free to use.

To forward information (a bit, a voice sample, a message, a cell, a packet or a frame) from a fixed or mobile input ter­minal (address, channel, line or trunk) to a selected output terminal (address, channel, line or trunk) of an exchange where the route selection or forwarding is controlled by local address information (based on the selected path) is called switching.

The telephone exchange will use network topology or routing information to prepare a connection. Hence, rout­ing precedes the establishment of a full circuit switched connection and is done only once while switching is done for each voice sample (using traditional circuit switching) during a connection.

To guarantee that all resources needed to enable com­munication in real time are available, they can be reserved by setting up a circuit connection along a route between the circuit end points. Information (voice and data) to be exchanged between such end points in a network can then be sent along the circuit connection that is routed and set up before the actual information exchange starts and then released when the information exchange ends. This guar­antees that all resources needed to enable voice and data communication in real time are reserved beforehand.

For historical reasons, circuit connection techniques are often associated with the synchronous transfer mode (STM) and pulse code modulation (PCM) using a 125 Ds frame rate for transmission of byte (8-bit) encoded voice samples.

Another technique, called asynchronous transfer mode (ATM), uses small packets called cells with a given fixed size (53 bytes) divided into a small header (5 bytes) and payload (48 bytes). The header does in this case not contain the address but rather path and channel identifiers, where a path is a bundle of channels. Routing is done in a setup mode where the destination address is sent as payload in a cell (hence used to carry control information). The traffic can then be switched at the path or channel level cell by cell in each node along the paths and channels routed and set up in advance from sources to destinations.

A third technique usually used for data communication is to send information in packets of a reasonable size, e. g. as short Internet Protocol (IP) packets, add addressing in­formation and other control information to the packet, and send it via a route toward its destination address. Each node on the way to the destination participates in the for­warding of the packet. For information that needs to be divided into several packets, forwarding is done for each packet in each node. To get deterministic packet transport delays circuit like reserved routes can be established using multi protocol label switching (MPLS) techniques.

Добавить комментарий

Ваш e-mail не будет опубликован. Обязательные поля помечены *