Abstract
This paper presents a multiprocessor architecture for the parallel implementation of the MultiStream Protocol, MSP, a new feature-rich transport protocol. The parallelism of the protocol implementation is based on protocol functions such as error correction, error notification, etc. By dividing the processing based on protocol functionality, the architecture not only achieves high throughput, but reduces the end-to-end processing latency of the protocol as well. In addition, the variation, or jitter, in the interpacket arrival times at the protocol receiver is stabilized. Analysis of a software implementation of MSP using the architecture presented in this paper demonstrates that if 10-MIPS processors are used to implement the protocol along with practical hardware support, the architecture may support over 1 Gbps of throughput at the transport layer.
Original language | English (US) |
---|---|
Title of host publication | IFIP Transactions C |
Subtitle of host publication | Communication Systems |
Editors | A. Danthine, O. Spaniol |
Publisher | Publ by Elsevier Science Publishers B.V. |
Pages | 135-150 |
Number of pages | 16 |
Edition | C-14 |
ISBN (Print) | 0444814817 |
State | Published - Dec 1 1993 |
Event | Proceedings of the IFIP TC6/WG6.4 4th International Conference on High Performance Networking - Liege, Belg Duration: Dec 14 1992 → Dec 18 1992 |
Other
Other | Proceedings of the IFIP TC6/WG6.4 4th International Conference on High Performance Networking |
---|---|
City | Liege, Belg |
Period | 12/14/92 → 12/18/92 |
All Science Journal Classification (ASJC) codes
- General Engineering