Teletraffic engineering/Routing

Module 15 of the Teletraffic Textbook

Summary
Routing (also spelt routeing) can be loosely described as the process of getting from here to there. Routing may be discussed in the context of telephone networks or computer networks. In telephone networks, routing is facilitated by switches in the network, whereby in computer networks routing is performed by routers in the network.
 * This article discusses routing in the context of telephone networks. For routing in other contexts, see routing in computer networks.

Definition: Routing in telephone networks
Routing in the context of telephone networks is the selection of a specific circiut group, for a given call or traffic stream, at an exchange in the network [4]. "The objective of routing is to establish a successful connection between any two exchangesin the network" [4]. By selecting routes that meet the constraints set by the user traffic and the network, routing determines which network resources (circuit group) should be used to transport which user traffic [1].

Different networks employ different routing techniques, but all communication networks share a basic routing functionality based on three core routing functions [1]:
 * 1) Assembling and distributing information on the state of the network and user traffic that is used to generate and select routes.
 * 2) Generating and selecting feasible and optimal routes based on network and user traffic state information.
 * 3) Forwarding user traffic along the selected routes.

The public switched telephone network (PSTN) architecture is made up of a hierarchy of exchanges (e.g local and regoinal exchanges) with each level of the hierarchy performing different functions [4]. Two adjacent exchanges in the network may be connected by several direct routes consisting of one or more circuits [4] as shown in figure 1.

In circuit-switched networks, such as the PSTN, switching and transmission resources are dedicated to a call along the path from source to destination for the complete duration of the call. Routing decisions are imperative in facilitating this process as they determine the most efficient links to use to connect users for a call [2]. Routing in the PSTN is done using a hop-by-hop approach [5]. When a user wants to make a call, they dial the destination number to which the call should be routed. This destination number is made up of a prefix (area code or national destination network), which identifies the geographical location of the called party, and a unique number (the subscriber number) linked to the prefix that identifies the exact destination to which the call should be routed [2,5]. The end exchange to which the calling party is connected (the originating exchange) uses the area code to identify the outgoing circuit group connecting to the first choice adjacent exchange en-route [5]. This circuit group is called the first choice route and is obtained using a routing table at the originating switch [5]. The function of the switch at the originating end exchange is to connect the switch input port to which the calling user is connected to a free outgoing circuit group in the first choice group [5]. If all the circuits along the first choice route are fully occupied, the switch then attempts to use an alternative route circuit group to route the call to the destination exchange [5]. The originating exchange then forwards the address to the adjacent exchange (first choice or alternate route), and the procedure is repeated at the adjacent exchange in order to reach the destination end exchange to which the called party is connected [5]. When the address reaches the destination exchange, it only needs to process the last part of the address to identify the switch input port that the called party is connected to [5].

Routing directs forwarding [3]. Forwarding of traffic can be done using connection-oriented or connectionless approaches [1]. In connection-oriented forwrding, forwarding instructions are installed in all the switches along a designated route before the route can be used to transport traffic [1]. Traffic forwarded using the connectionless approach carries its own forwarding information either as precise routing commands for each switch along a route or as hints that may be autonomously interpreted by any switch in the network [1].

In PSTN, forwarding of traffic is based on the connection-oriented approach. Call routing is achieved using pre-computed routing tables, containing all the possible pre-defined routes for a connection, at each switch [2]. The pre-defined routes specified in the routing table include information of a direct route (or routes) to be used under normal traffic and network conditions (e.g no link failure or network congestion) as well as alternative routes that should be used in the event that all circuits along the direct route are fully occupied [2]. An alternative route may be an indirect route consisting of several circuit groups connecting two exchanges via other exchanges [4]. The following example illustrates the use of an alternative route to connect two exchanges in the event of the direct route being congetsed.

Example
Let us suppose we have three areas served by exchanges A, B and C as shown in Figure 2. At a grade of service of 0.002 (blocking probability of 0.2%), using an Erlang B table generator, we would require 81 circuits to carry 60 erlangs of traffic during the busy hour. If we decreased the number of circuits between exchanges B and C while still maintaining the busy hour traffic at 60 erlangs, we would be increasing the efficiency of the B-C link at the expense of a reduced grade of service (or increased blocking probability). This would result in bocking along the B-C route. The switch at B would therefore have to route traffic meant for C that experienced congestion along the direct B-C route via exchange A using the B-A route. The switch at A would then route this traffic along the A-C route to exchange C.

Routing and flow control are network control processes that operate in a dynamic environment determined by the network and user traffic state [1]. Routing schemes may be classified as static or dynamic depending on the types of state changes they respond to, as well as the speed of their responses [1]. Possible state changes that occur include link and switch failure, variations in user traffic, and availability of network resources [1]. In static routing systems, routing is based on the expected, as opposed to the actual, user and network behaviour and does not change according to the current state of the newtork and user traffic [1]. In dynamic routing systems, the routing pattern is automatically altered based on present changes in the user and network state [1]. Cost and performance tradeoffs are normally required when designing a routing system for specific networking environments [1].

Exercises

 * 1) What is routing in the context of telephone networks? Click here for answer
 * 2) What are the two types of routing schemes and how do they differ? Click here for answer
 * 3) What is the function of a switch and how does it aid routing in a telephone network? Click here for answer