An application of fuzzy hypothesis testing to signal integration

A method of fuzzy likelihood-ratio hypothesis testing is applied to binary signal integration. Imprecise parameters of the distribution under each hypothesis are modeled as fuzzy numbers. The false alarm rate and detection rate are therefore also imprecise. The effects of parameter imprecision on integration performance are examined using an example; the method is applied to envelope detection and integration of fixed amplitude signals in additive white Gaussian noise. For low parameter imprecision, the integration system produces performance which is nearly crisp. However, as parameter imprecision increases significantly, the performance quickly becomes so imprecise that the system may no longer be of practical use