Limitations, Trade-offs and Recommendations for Source Qualification and Combining Algorithms
Douglas Arnold, Meinberg USA
Location: Beacon B
Date/Time: Thursday, Jan. 25, 8:57 a.m.
For a timing equipment to be resilient timing source inputs must be qualified. Qualification mechanisms include checking that received metadata attributes are in expected ranges. The behavior of the received timing input should also be compared to the local clock to determine if there are unexpected jumps or an unreasonably large time drift between the input and the local clock. High resilience timing equipment will generally make use of multiple timing inputs so that bad inputs can be identified by comparison with other inputs. Such techniques must be flexible enough to reliably and quickly detect bad input signals in the presence of noisy data, but also not give false positives that lead to nuisance alarms for network operators.
This presentation will outline the fundamental tradeoffs and limitations in timing input source qualification and selection and illustrate them with examples. The subject of this presentation will not be to promote a specific algorithm or compare popular algorithms, but rather to describe the inherent trade-offs that should be considered when algorithms are designed. For example, the need to filter noisy data versus the need to avoid false positives. In addition, a set of recommended principles for source qualification and selection will be presented.