The Use of the GLRT for Revealing Faults in Atomic Frequency Standards

Abstract

Atomic clocks are instruments widely employed in many synchronization systems. When such measurement instruments are inserted in complex systems as telecommunication networks, global satellite navigation systems, tests of fundamental physics of matter, an unexpected and anomalous behavior or a degradation of the performance of the clock may give rise to an error condition in the global system (Galysh et al., 1996); (Vioarsson et al., 2000). In some cases, as in the GPS system, the anomaly must be identified and a real-time alarm signal must be transmitted to user. Aerospace systems and navigation support systems used in the area of personal security, for both military and civil purposes, must satisfy strict requirements of the parameters relative to the integrity, reliability, availability and accuracy of the signal. The possible lacking of information concerning one of the above features, may imply a series of inefficiencies and system bugs for end-users. Monitoring stability of atomic clock frequency data is important for guaranteeing the correct behavior of the electronic system where they are inserted. The principal application where the frequency stability monitoring is a challenging problem is in GNSS systems (like GPS or Galileo) where the overall system performance critically depends on performance of on– board clocks. When the clock behaves bad, thus the anomaly has to be detected fast in order to provide an adequate action for restoring the correct behavior of the clock. In the field of navigation system “integrity”, most of the studies are related to satellite integrity (Bruce et al., 2000) and not specifically to that of the embedded clock, while recently a study on GPS clock integrity showed GPS clock strange behaviors (Weiss et al., 2006), asking for suitable new statistical tools for its characterization. The scientific and industrial community has done a lot of efforts for the theoretical characterization of the behavior of atomic clocks. The purpose is to improve the accuracy and reliability of these instruments while reducing their size and cost. Since such objectives are often in contradiction, the scientific and industrial research is investigating innovative techniques to overcome the limits imposed by the technological development. A common assumption when analyzing atomic clock data is that clock noise is stationary or that at least increments in the frequency values are stationary and thus data are examined by using stability analysis tools such as the Allan variance (IEEEstd, 1999); (D. W. Allan, 1987). The scientific literature has shown how this hypothesis cannot be always verified in reality, in particular in the application context of the satellite navigation or in experiments of fundamental physics of matter. If this hypothesis is not satisfied, the accuracy and reliability 2