Abstract

A study was made of the flow of data and the simultaneous processing needed to reduce the 10/sup 7/ to 10/sup 8/ triggers per second expected at ISABELLE to a number of events on the order of 10 to 100 per second which would be written on magnetic tape. It was assumed that within 100 ns of the event a fast pretrigger would have reduced the data rate to at most 10/sup 7/ per second. At that point, data from all sense elements in the experiment would be fed into a 1-..mu..s-long pipeline. Within the first 1 ..mu..s (while the data are in the first pipeline) another level of triggering would reduce the trigger rate to at most 10/sup 6/ per second. The data would then be fed into a second pipeline which is 50 ..mu..s long. During the 50 ..mu..s the data are in the second pipeline, a more sophisticated level of triggering (slow trigger) would reduce the trigger rate to a level that can be handled by standard data processing techniques (microprocessors or larger machines), i.e., 10/sup 2/ to 10/sup 3/ per second. The pipelines and the buffer between them, a sequential address memory, are described first, andmore » then several alternative schemes for the pretrigger and slow trigger are presented. 10 figures.« le Document type: Report

Abstract

International audience; Performance metrics in the air traffic management (ATM) community are becoming a strategic issue, and they are getting more and more attention. However, defining such metrics is a difficult problem. In this paper, we show how arithmetic simulations can be used to give performance information. We also point out that many different metrics can be defined, each of them giving different results regarding efficiency. We conclude that an extreme caution must be applied when interpreting results.