Performance of ATM over Satellite


Outline of Research

The Communications Group of the Faculty of Computer Science,
University of New Brunswick, Fredericton, NB, Canada is 
currently involved in the evaluation and application of ATM 
technologies, particularly over satellite. Two projects are being 
pursued at this time.

	Participating in this research and development work are 
(in alphabetical order) John M. Dedourek (Professor), Dr. Bernd 
J. Kurz (Professor), Adam Reece (MCS graduate student). 

	This work is carried out under contract and with the
collaboration of Cygnus Technology Ltd., Fredericton, N.B., Fundy 
Cable Ltd., Fredericton, N.B. and Telesat Canada, Ottawa.

Project 1

In partnership with two commercial enterprises and Telesat Canada we have obtained a satellite groundstation in Fredericton which connects us over T1/T3 speed links to the ATM-based OCRINet in Ottawa, Canada over the ANIK-2E satellite. Apart from educational use for occasional teleconferencing and seminar distribution, we are using this link to collect ATM cell error statistics in a production environment. This data will be used to establish a useful channel model, including data unit error rates. We are interested in measuring not only simple (average) cell errors, but also distributions and correlations of such errors (expressed in various error types such as cell error, cell burst error, long cell burst error, and link drop error statistics). All these error types are necessary in order to arrive at reasonable performance results via our simulation protocol testbed. Unfortunately, only the ATM cell loss rate has been generally used in published work for ATM simulation, which according to our experience is insufficient to obtain useful simulation results for ATM-based protocol suites because segmentation/reassembly effects require the above error types for the protocol layers above ATM. At this time we have available initial simple ATM cell error statistics, but are still in the process of completing this work.

Project 2

A research contract, funded by a local communications provider and IRAP (a government funded Industrial Research Assistant Program), is being carried out to develop an interface of HDLC over ATM. This interface, that probably uses AAL5 has not yet been standardized by the ATM Forum. The HDLC link layer is to guarantee reliable link communication, and was required by our contractor for use in legacy networks (to run over ATM directly). We have defined a mapping from HDLC frames to AAL5 PDUs, currently for point- to-point operations. The possibility of operating packet- -switched networks under X.25 over ATM directly will be opened as well with the availability of a HDLC/AAL5 interface. Similarly, our new AAL5 mapping can form the basis for the operation of SNA networks employing IBM's SDLC protocols directly over ATM. The important issues related to address mapping and signalling will be addressed in a later phase. Carrying existing network data directly over new ATM networks appears to attract much corporate interest because it combines the best of the past and the future. This approach allows the preservation of large investments made by corporations into legacy networking operations (X.25, SNA, etc.) and benefits from the advantages of modern ATM networks to carry efficiently many traffic streams with opposing service requirements. Until that time when the much-promised multimedia traffic will come into existence to pay the bills of ATM networks, the legacy networks might just come to the rescue of ATM, and fully profit from the new universal ATM networks. Project Status: Started: January 1996 Completed: September 1996


Back to: Kurz's Research Page | Kurz's Home Page

Last revised: 6 June 1997 by Bernd Kurz bjkurz@unb.ca