Multi-rate Protocol Testing

As data services and video conferencing became more popular, the need arose for bandwidth on demand. The issue facing ISDN in general was that bandwidth wasn't flexible. You had a 23B+D 1.544 Mbits/sec PRI span but only one 64 Kbits/sec B-Channel could be used per call.

A normal voice call was fine on a single beared channel but a video conference might require the combined bandwidth of multiple simultaneous B-Channels all the way up to an entire span. A means was needed to dynamically allocate bandwidth. Multiple B-Channels could be nailed up but nailed connections represent permanent bandwidth while the need was for dynamically allocated bandwidth.

For example, bandwidth for an occasional video call at H0 rates of 384 K bits/sec requiring 4 simultaneous B-Channels could not be dynamically allocated on a span. Four nailed B-Channels could be used and the video conference equipment use various protocols to transmit data through the "pipes" but unless those four "pipes" were in continuous use, they became highly cost ineffective.

Similarly, supporting a crisp, management quality videoconference might require the entire 1.544 mb/sec bandwidth of a T1 span. Either an expensive T1 span would have to be dedicated for this purpose, or staff would have to be on hand to free up necessary bandwidth which might mean forcibly terminating existing calls.

Dynamic Channel Bonding

By dynamically bonding two or more B-Channels, bandwidth could be allocated as needed. A request for bandwidth was just that, a request. If the PBX could not grant the request, a lesser amount would be offered. A calling device might request 8 B-Channels when only 4 were available. It was up to the requestor to accept the lesser bandwidth offer or not.

Once a connection was established, data was packetized and sent over the bonded B-Channels. To insure data was correctly reassembled on the far end, a technique called timeslot sequence integrity was used. In circuit switched equipment, only one timeslot can be active at any given time. So, if four B-channels were bonded together, four timeslots would be allocated and in use but only one timeslot would be active at any given moment. As long as both sides knew which timeslots were part of the connection, and the sequence in which they became active was known, transmitted data could be correctly reassembled on the far end.

With the implementation of non-facility associated signalling, multi-megabit bandwidth became dynamically available.

Testing

Feature level testing while broad in scope was fairly straight forward in that data corruption was immediately detectable. From a feature test standpoint, the only thing different from testing any other piece of new terminal equipment was the bandwidth negotiation and managing the channels. Because multiple channels tied to a single connection were involved, care had to be taken to verify billing record accuracy. Protocol level changes were required and had to be tested.

Ultimately, dynamic bandwidth allocation was an improvement but may have come too late for ISDN as video conference had already designed ways around the problem and had little motivation to change designs absent major price improvements in ISDN by the telco's which never happened.

For high end videoconferencing applications, systems relying on ISDN as the carrier are still widely used.