hes@ccvr1.cc.ncsu.edu (Henry E. Schaffer) (10/02/90)
In article <12433@accuvax.nwu.edu> dave@westmark.westmark.com (Dave Levenson) writes: X-Telecom-Digest: Volume 10, Issue 666, Message 6 of 11 >Future trends go toward >allocating only the bandwidth required to every connection. Rather >than assign 64kbit/second of bandwidth to every conversation, whether >or not it needs it, the future network will only assign the bandwidth >actually required by the message channel being carried. Speech >compression and coding technology has advanced a long way since the >first digital telephony standards were written. While processing is getting cheaper, it still requires extra equipment and therefore there will be a tradeoff depending on the relative costs of this processing and the savings in bandwidth. Bandwidth seems to be getting cheaper, and we can get a very rough indication of the costs of bandwidth vs. electronics from looking at the relative costs of different amounts of bandwidth, e.g., 64 kbs vs. T1. The 24 fold increase in bandwidth typically costs about 5 times as much. Even assuming that T1 termination electronics cost no more than 64 kbs, this suggests that under a fifth of the total line costs are attributable to bandwidth. (I'm also assuming that the charges are related to the costs of providing the service.) I would expect there to be progress in dynamic allocation of bandwidth, and the ability to request it, but if the cost goes up slowly with the extra bandwidth then there would not be much pressure to do processing to minimize voice bandwidth except for the most expensive lines such as transoceanic ones. henry schaffer n c state univ