We consider nonparametric or universal sequential hypothesis testing problemwhen the distribution under the null hypothesis is fully known but thealternate hypothesis corresponds to some other unknown distribution. Thesealgorithms are primarily motivated from spectrum sensing in Cognitive Radiosand intruder detection in wireless sensor networks. We use easily implementableuniversal lossless source codes to propose simple algorithms for such a setup.The algorithms are first proposed for discrete alphabet. Their performance andasymptotic properties are studied theoretically. Later these are extended tocontinuous alphabets. Their performance with two well known universal sourcecodes, Lempel-Ziv code and Krichevsky-Trofimov estimator with ArithmeticEncoder are compared. These algorithms are also compared with the tests usingvarious other nonparametric estimators. Finally a decentralized versionutilizing spatial diversity is also proposed. Its performance is analysed andasymptotic properties are proved.
展开▼
