...\author

Olivetti Research Ltd., Trumpington Street, Cambridge CB2 1QA, United Kingdom

...\author

Cambridge University Computer Laboratory, Corn Exchange Street, Cambridge CB2 3QG, United Kingdom

...

Such matters are described in more detail in a previous article in IEEE Network [15].

...

This adaptation layer, called UDL, places 8 bits of sequencing information in the header of each cell, and is the one used in the performance example presented in Section 6.2.

...

The December 1991 CCITT draft I.361 extended the payload type to three bits and defined this end-to-end user data bit. It is termed the `ATM layer user-user indication'.

...

Our links are currently 100 Mbit/s and of course full duplex.

...

In MSSAR (and AAL-5), MIDs are not used. Further in MSNL there is no a priori division of the header into separate VCI and VPI fields (c.f. Section 4).

...

Adaptation protocols based on a header flag, such as the one described in Section 4 do not require adaptation layer headers in every cell payload.

...

To avoid this duplication in practice, the transport protocol must be able to accept a `data-ok' indication from the lower layers.

...

Although CCITT has defined that B-ISDN should not reorder cells, we are considering all types of ATM system.

...

Assuming a 32 bit CRC, there are twelve 32 bit words in each cell payload.

...

The major between the two machines is that the /200 has an 800 Mbit/s Turbochannel while the /25 has a 400 Mbit/s version.

...

The B-ISDN switch manufacturers are faced with the same VCI/VPI look-up problem, so are probably able easily to control VCI space. The question is whether they will make VCI non-sparseness guarantees to the customer.