Adaptive detection with time series models

Adaptive detectors based on time series models can yield accurate detection algorithms, if an appropriate model order and model type is used. Using models of a model order that is either too high or too low will result in reduced detection performance. Statistical order selection offers a practical solution for the adaptive selection of a model order from data. With the combined information criterion CIC as an order selection criterion an optimal trade-off of underfit and overfit is made. An adaptive detection algorithm has been developed that is based on this selected model. This detector has been compared to detectors based on fixed order models and detectors based on the periodogram in a simulation study. Also, the new detector has been applied to experimental data.