Novel algorithms for distributed sequential hypothesis testing

Abstract

This paper considers sequential hypothesis testing in a decentralized framework. We start with two simple decentralized sequential hypothesis testing algorithms. One of which is later proved to be asymptotically Bayes optimal. We also consider composite versions of decentralized sequential hypothesis testing. A novel nonparametric version for decentralized sequential hypothesis testing using universal source coding theory is developed. Finally we design a simple decentralized multihypothesis sequential detection algorithm.

Novel distributed sequential nonparametric tests for spectrum sensing

“…This has also been used in [7]; the main difference is in the detection algorithms used at the local nodes. Due to non availability of modulation/coding and channel gains of the PU, we use nonparametric algorithms at the local SUs.…”

“…Quickest change detection problem literature is summarized in [6]. Distributed sequential detection has been studied in [7], [8], [9] and [10]. The problem considered in [9] and [10] is of sequential distributed detection of two simple hypothesis when the distribution under both hypothesis is known.…”

“…In [7] and [8] Fusion Center (FC) noise is considered but handled very differently from [11] via physical layer fusion. The approach of [7] and [8] has an additional advantage that it requires a single low bandwidth reporting channel with minimal transmission delays. Our present algorithm differs from [7] and [8] in the local algorithms used by the CR nodes.…”

“…The approach of [7] and [8] has an additional advantage that it requires a single low bandwidth reporting channel with minimal transmission delays. Our present algorithm differs from [7] and [8] in the local algorithms used by the CR nodes. We show that these algorithms can perform better than the algorithm in [8] when the spectrum sensing problem can be formulated as a nonparametric mean change problem.…”

“…We show that these algorithms can perform better than the algorithm in [8] when the spectrum sensing problem can be formulated as a nonparametric mean change problem. Also the present algorithms are completely nonparametric while the algorithms in [7] and [8] assume that the distribution under H0 is known.…”