TE ECOMMUNICATIONS NETWORKS
The present invention relates to telecommunications networks, and in particular to subscriber connections to the internet. DESCRIPTION OF THE RELATING ART
Subscribers are able to connect to the internet by way of the PSTN/ISDN network. However, one of the problems associated with such public network connections is that the switching resources are often not used efficiently while the subscriber is connected to the internet. This is because the total bandwidth of the subscriber lines is generally much higher than used bandwidth when connected to the internet, due to the bursty characteristics of internet traffic. Thus the internet connection is used inefficiently. Furthermore, since the connection time for subscribers connected to the internet is generally much longer than for normal connections, the public switch apparatus will probably experience problems with congestion. Such a congestion problem is likely only to increase as subscriber use increases.
One previously considered solution to this problem is to use more switching resources between the subscribers and the internet connection. An alternative system inserts switching equipment between the subscribers and the public switching equipment . A further solution would be to put additional switching equipment in the actual subscriber stage .
However the problem with all of these solutions is that they require installation of new hardware. SUMMARY OF THE INVENTION
Thus it is an object of the present invention to provide an efficient switching system for enabling subscribers to connect to the internet via a public telecommunications system.
According to the present invention there is
provided a method of controlling the connection of a subscriber to a network in which the connection is maintained only during data transfer between the subscriber and the network. According to a second aspect of the present invention, there is provided a node apparatus for connecting a subscriber to a network, the apparatus including a control means which is operable to maintain a subscriber's connection to a network only while data is transferred between the network and the subscriber. BRIEF DESCRIPTION OF THE DRAWINGS
Figure 1 is a block diagram of a previously considered network connection apparatus;
Figure 2 is a block diagram of further previously considered network connections apparatus;
Figure 3, 4 and 5 are block diagrams of a network connection apparatus embodying the present invention during respective connection processes; and
Figure 6 is a flowchart illustrating steps for operating the network connection apparatus shown in Figures 3, 4 and 5. DETAILED DESCRIPTION OF THE DRAWINGS
Figure 1 shows a previously considered switching apparatus in which subscriber lines 10 are connected via a public switch 12 and an ISDN Primary Rate
Interface (PRI) 14 to terminating equipment 16, often referred to as an Internet Access Server (IAS) . The IAS connects to an internet connection 18. A subscriber who wants to be connected to the internet dials a number which identifies the access where the IAS 16 exists. Normally this equipment is connected via the PRI 14, although it could be integrated into the public switch.
The IAS 16 terminates the link level normally PPP, and handles authorisation, etc. It will also terminate modem signalling if the subscriber has an analogue
line .
The total bandwidth of the subscriber lines is much higher than the actual traffic generated by the applications which the subscriber is running, since the traffic is bursty, and so the Figure 1 connection apparatus has poor economy. Furthermore, since the connection time for internet-connected subscribers is generally much longer than for normal connections, the public switch and network experiences congestion problems. These problems can only increase when the penetration of subscribers into the internet becomes much higher than today.
The congestion could occur within the switch, or remote subscriber stage, or in the lines connecting that stage to the switch, or in the network if the IAS 16 is connected to another switch in the public network.
Figure 2 illustrates three possible solutions to overcome the problems of the Figure 1 switch connection apparatus.
A first solution is to connect more switching resources 20 within the public switch 12.
Alternatively, additional switching equipment 22 can be inserted between the subscriber lines 10 and the switch 12. An optional connection 24 can be provided between the switches 22 in order to bypass the public switch 12.
A third solution would be to include additional switching 26 in the subscriber stage, which provides an optional connection 28 to the terminal equipment 16. As mentioned previously, the problem with all of these solutions is that they all require new installations, which leads to more hardware being required. A preferred embodiment of the present invention will now be described with reference to Figures 3 to 6 ,
Figures 3, 4 and 5 being block diagrams illustrating the connection apparatus in different respective stages, and Figure 6 being a flowchart illustrating those stages. A subscriber uses a normal set up procedure for setting up a connection to the switch. The present invention is concerned with how the connection to the internet is handled in the public switching apparatus. Internet connection traffic to and from a subscriber tends to be "bursty", meaning that the data transfer is carried out in relatively short bursts. The bursty nature of the data traffic transfer means that the connection resources are somewhat underused. Thus, in the connection system embodying the present invention, instead of a subscriber being online with the IAS constantly, a scanning process takes place. The subscriber is then connected to the IAS only intermittently, when it is determined that traffic is to be transferred between the subscriber and the IAS.
As shown in Figures 3, 4 and 5, a number of subscribers 10 are connected to a local exchange 30. The local exchange 30 includes a control unit 32 which operates to control a switching sub-stage 34 and a main switching stage 36. The local exchange is connected to the IAS 40 by way of a PRI 38, 39. For the sake of clarity, only 3 B-channels (Bl, B2 and B3) are shown in Figures 3, 4 and 5. In addition, the D-channel 39 (A, D) is shown. The D-channel connections 39 enable control data to be transferred between the IAS 40 and the control unit 32 of the exchange 30. The PRI actually is only one interface, but is illustrated schematically in Figures 3, 4 and 5. In the USA, for example, the interface has 23B and ID channels, and in European 30B and ID channels (called primary rate access PRA) .
Operation of the connection equipment of Figures 3, 4 and 5 will now be described with reference to those Figures and to the flowchart shown in Figure 6. At step A of Figure 6, the IAS 40 signals the control unit 32 on the control channel 39 in order to find out which B-channels are currently free for use, i.e. they are idle .
Subscribers who will be offered channels are identified using a standard ISDN call reference. This call reference is associated with a specific idle channel by the IAS 40.
At step B, when the control unit 32 receives the control signal, it will connect the spare B-channels to the subscribers indicated in associated call references. Appropriate paths are connected within the sub-stage switch 34 and the main switching unit 36 (step C) and an acknowledgement is returned to the IAS 40 (step D) .
When the IAS receives the acknowledgement from the control unit 32, it sends out in step E a
"query_for_connection" message to the subscriber terminals. This is done in band within the different B-channels. The query_for_connection message is used to interrogate the connected subscribers to ascertain whether they have internet protocol (IP) datagrams to be sent to the IAS 40. Although any scanning frequency can be used, a desirable level would be 4 times per second.
Alternatively, if there are IP datagrams to be transmitted from the IAS 40 to a subscriber, a
"connect_request" message will be sent in step F instead of the query_for_connection message.
In step G, following step E, IP datagrams are sent from the subscriber to the network, after that subscriber has seized the connection for its own use. The IP datagrams are transmitted to the IAS 40 and an
internal timer T± is started.
The subscribers which receive a connect_request message in step F respond with a "connect_accept" message (step H) so that they can receive IP datagrams from the IAS 40.
The IAS then records that a B-channel has been seized by the subscriber.
Every time an IP datagram is sent, the timer T1 is restarted. If the timer Tx times out, no more IP datagrams are expected to be sent and so the subscriber terminal will release the connection by sending a "release_request" message (in band) and starting a timer T2 (step I in Figure 6) . If the subscriber terminal has not received a "query_for_connection" message within time T2 then a new release request will be sent and T2 restarted. When the IAS receives the release_request message, it marks the corresponding B- channel as now being idle and so that B-channel can be included in the list of B-channels for scanning. The newly-used B-channels are noted as being busy
(step J) and will not be used in the next scan. The process then continues from step A where the currently idle subscriber channels are used for new subscribers connections . One possible additional feature would be to have forced release of a subscriber terminal. This would ensure better distribution and use of resources at the IAS. For example, if there are more than x subscriber channels occupied and some of the terminals have been using a connection for more than y seconds, then the IAS will send a forced release message to those terminals. The values of x and y can be set depending on the scan frequency concentration factor and type of flow. The protocol considerations for the present application must not effect the TCP/IP layers, and
therefore the protocol will be below the IP layer. It could possibly be an extension to the PPP layer.
The mechanism described above might work for analogue accesses, but then resynchronisation will be needed every time there is a seizure. Furthermore, new types of modems will need to be used.
For ISDN accesses (primarily the basic rate access) this solution will work well. When a B-channel in a basic rate access is not used, both the user and the network will send binary ones which will fit very well in the HDLC structure of PPP.
The invention is described herein with reference to a situation in which the internet IAS resides outside the local switch. However, the invention may also be used in a generally similar way when the internet IAS is within the local switch. In that event, the control signalling sent between the IAS and switch can be made more efficient .
In general, in using the invention, the subscriber might perhaps be charged not for the total time over which his connection is active, but instead for the actual time for which he has a connection to the internet IAS. Alternatively, the charge might be calculated on the basis of the number of bits of data transmitted.
For subscribers which do not utilize the scanning procedures, a normal continuous on-line connection can be established.
There are thus disclosed a method and an apparatus which allow efficient usage of available equipment.