Methods and Tools for Frequency Jump Detection

The prompt detection of jumps in frequency data is a topic currently receiving large attention from both the scientific and industrial communities. In fact, the frequency stability of the clock inserted in advanced telecommunication systems for synchronization purposes, has to be continuously monitored in order guarantee the target accuracy of the whole system. Many research and industrial laboratories analyze anomalies in frequency data by using algorithms mainly based on the Allan variance estimate of available data. Recently, a fault detection technique based on statistical tools, the Generalized Likelihood Ratio Test (GLRT), has been proposed effectively for revealing anomalies by means of the real–time processing of acquired data. This technique can be employed for emitting an alarm signal while the data acquisition process is going on and for having a good estimate of the jump location. In this paper, a performance comparison between the GLRT and other two algorithms (called BLKAVG and SEQAVG) usually employed in scientific labs for analyzing frequency jumps, is presented. The goal of the comparison process is the validation of the GLRT as a novel technique for revealing effectively anomalies occurrences in frequency data. In particular, it is shown that information given by GLRT on data parameters are complementary to those given by BLKAVG and SEQAVG.