To ensure high accuracy results from an integrated GPS/INS system, the carrier phase observables have to be used to update the filter's states. As a prerequisite the integer ambiguities must be resolved before using carrier phase measurements. However, a cycle slip that remains undetected (and uncorrected) will significantly degrade the filter's performance. In this paper, an algorithm that can effectively detect and identify any type of cycle slip is presented. The algorithm uses additional information provided by the INS, and applies a statistical technique known as the cumulative-sum (CUSUM) test. In this approach, cycle slip decision values can be computed from the INS-predicted position (due to the fact that its short-term accuracy is very high), and the CUSUM test used to detect cycle slips (as it is very sensitive to abrupt changes of mean values). Test results are presented to demonstrate the effectiveness of the proposed algorithm.

We develop a test of evolutionary change that incorporates a null hypothesis of homogeneity, which encompasses time invariance in the variance and autocovariance structure of residuals from estimated econometric relationships. The test framework is based on examining whether shifts in spectral decomposition between two frames of data are significant. Rejection of the null hypothesis will point not only to weak nonstationarity but to shifts in the structure of the second-order moments of the limiting distribution of the random process. This would indicate that the second-order properties of any underlying attractor set has changed in a statistically significant way, pointing to the presence of evolutionary change. A demonstration of the test's applicability to a real-world macroeconomic problem is accomplished by applying the test to the Australian Building Society Deposits (ABSD) model.

We investigate the size and power properties of the AH test of evolutionary change. This involves examining whether the size results are sensitive to both the number of individual frequencies estimated and the spectral shape adopted under the null hypothesis. The power tests examine whether the test has good power to detect shifts in both spectral position (variance) and spectral shape (autocovariance structure).