New access technology of N-ISDN Ao/di

With the rapid development of the Internet, in the WAN range, people increasingly use N-ISDN remote access, but the early N-ISDN protocol design did not expect to use it for Internet access, this access to the Internet and in the way of modem use circuit switched network dial-up connection, The disadvantage is wasting bandwidth.

Users must first establish a circuit connection when surfing the Internet, while users browsing, waiting for Web pages, sending and receiving e-mail, exclusive B channel is not fully utilized, which is a waste for ISPs, users and telephone companies. In order to improve the use efficiency of the network, a new technology ao/di.

Ao/di (Always on/dynamic ISDN) is a new access technology proposed by via (Vendor\ ' ISDN Association) in 1996, Its core idea is to use the packet switching network (X.25 network) and ISDN D-Channel to transfer small data volume of the Internet business, such as news updates, e-mail send and receive, And at any time to maintain a X.25 SVC (Exchange Virtual Circuit) connection, when the data volume is too large to exceed the D-channel capacity, using BAP/BACP/MLPPP and other protocols to automatically allocate B channel to adapt to large data volume business, such as downloading files, pictures and even real-time voice and video, The B channel is automatically freed when the peak of data volume is past. The process of allocating and releasing B channels is transparent to the user. Because the SVC connection always exists, the user does not have to redial every time it is received, and the ISP can also send data at any time (such as notification of new mail), which is what always on means. The bearer B channel is dynamically allocated according to the best principle, thus increasing the resource utilization, which is the meaning of dynamic ISDN.

Operation of Ao/di

In general, modern switches have X.25 packet processing modules, BACP (Bandwidth allocation control Protocol) is the IETF RFC2125 protocol for PPP bandwidth allocation. In Ao/di, users use X.25 to call an ISP. The TCP/IP packet is routed through the MLPPP package in the D-Channel X.25 logical channel, and when required, the bearer channel provides standby bandwidth. When a hosting channel uses MLPPP, it does not use the q.922 and X.25 encapsulation as the D-Channel does. The circuit connection between the ISP and the user is established on the B channel, and the IP packet is transmitted directly through the MLPPP package on the B channel.

Although the X.25 protocol is inefficient, Ao/di uses it to take advantage of the common packet processing module on the current switch, and the X.25 packet connection on the D channel is the primary connection of the MLPPP. The reason for maintaining full-duplex is always active because Ao/di maintains physical layer (2b1q synchronous modulation) and q.922 data link Layer activation state even when there is no q.931 signaling information. and a small number of packet data, through the X.25 virtual call, between the ISP and the user transmission, this way also make full use of the packet processing module, avoid the use of switches, reduce the burden of the switch.

The general operation process is as follows: First, a X.25 SVC call from the user to the ISP, unless the PC is restarted, usually, this svc always exists. When a B-hosted channel is required, the user sends the appropriate phone number to the ISP, and the hosting channel connects to the ISP,MLPPP through the switch to the unified management of these B-channel and SVC connections.

Ao/di uses the BACP-controlled MLPPP protocol to negotiate bandwidth, manage phone number exchange, and combine subsequent added bandwidth. BACP/MLPPP allows ISPs and users to initiate or dismantle a B call to request more bandwidth, a symmetrical way in which users and ISPs can send large amounts of data to each other.

Ao/di has many improvements to the traditional PPP X.25 encapsulation: The physical layer uses SVC instead of the PVC call. The Cud field has a reserved value for Ao/di to use, and the DL layer encapsulates the PPP frame head directly in the X.25 frame without replacing it.

Because D-channel only 16KB/S,TCP/IP over X.25/D channel has limited application scope. When the D channel is not sufficient, use the BACP request to establish a B channel connection, this kind of statistical multiplex method uses the switch efficiently, is advantageous to the telephone company and the ISP. It is noteworthy that each MLPPP connection can send a BACP signaling to allocate the B channel. Even if the X.25 connection on the D channel does not exist, the Ao/di receiver can also handle the BACP signaling sent through the B channel, which is compatible with MLPPP and guarantees interoperability between the BACP-enabled device and the device that does not support BACP. Even if BACP cannot successfully negotiate with the peer, the X.25 call can remain connected to transfer the PPP load. The B-channel can be allocated or released even without the BAP negotiation process. The BAP negotiation process must have the following entities: The AO/DI server, which sends a BAP Callback Request to the client to initiate bandwidth allocation.

The Ao/di client can send BAP call request to the server to initiate bandwidth allocation and handle the BAP Callback Request sent by the server. In order to support the X.25 address at the same time, Ao/di recommended phone deltas number to send the method.

The composition of the Ao/di protocol heap

Ao/di protocol heap: TCP/IP through the MLPPP package can choose the D-Channel X.25 virtual connection or B-Channel circuit connection, generally only use the former. BAC is a bandwidth allocation controller that uses the BACP protocol to dynamically allocate or release B channels on a X.25 virtual connection; A dialer uses q.931 to establish a call.

Dynamically allocate and release bandwidth

In order to dynamically allocate bandwidth, it is necessary to monitor and estimate data traffic in real time, and based on these estimates, some triggering mechanisms can be set up to allocate the bearer channel at the appropriate time. The trigger allocates a B channel when the traffic is large until the D channel is transmitted continuously for 5 seconds or the pending data in the queue is greater than 7500 bytes. B-Channel establishment, negotiation, initialization takes 3 seconds, which means that the D channel can also transmit 4500 bytes of data within 3 seconds; After the B channel is established, the D channel is in an idle activation state (that is, the link remains, but no data is transferred). This is a highly efficient and error-prone approach. For compatibility purposes, however, Ao/di requires the receiver to process data from both the D channel (X.25) and the B channel at the same time.

There are several ways to implement a trigger, for example: set up the upper layer of software to put IP packets into a buffer, mlppp from the buffer to obtain IP packets, or by the X.25 through the D channel, or through the B channel to the ISP. MLPPP each IP packet is sent, it clears it from the buffer, setting the time to empty the buffer is T seconds:

-When t<5s, only use D channel, do not assign B channel;

-When 5s
If there is an available B-channel, increase the number of connections with MLPPP;

If no B channel is available, continue to monitor the buffer and monitor the B channel for availability;

-When 10s
If two B channels are available, increase the number of connections with MLPPP;

If a B channel is available, add a connection with MLPPP and monitor both the buffer and another B channel occupancy;

If no B channel is available, continue to monitor the buffer and monitor the B channel for availability;

When Ao/di allocates a B channel, it does not affect the B channel in which the call is being spoken.

Release B Channel:

When the B-channel is used to transmit the data, the channel must be released, BACP support such a request, but must take into account the requirements of the application, such as a page after downloading 5-10 seconds, users do not have the action, you can think that the user is carefully browsing the page, you can disconnect B channel. After sending and receiving the e-mail, you can disconnect the B channel. When the Ao/di occupies the B channel, sometimes it just hits the ISDN incoming call, or the other device on the s/t interface calls, then Ao/di should be able to give up B channel.

The network structure of Ao/di

Ao/di changes to the existing network structure, as shown in Figure 2, the user device D channel directly to the switch X.25 group to the ph (Packet Handler) packet processor, ph using the switching network with the ISP's equipment, B channel via TSI (time slot Exchange), The ISP is connected via the PRI interface or the BRI interface of ISDN, which is a circuit-switched network.

Although the Ao/di agreement is still being perfected, it has attracted more and more attention. Via organization has already launched a number of companies to support Ao/di end products and end products.