Distributed Detection of Correlated Random Processes under Energy and Bandwidth Constraints

JUAN AUGUSTO MAYA; LEONARDO REY VEGA; CECILIA GABRIELA GALARZA; ANDRÉS ALTIERI

Conferencia; 8th IEEE Sensor Array and Multichannel Signal Processing Workshop; 2014

IEEE

We analyze a binary hypothesis testing problembuilt on a wireless sensor network (WSN). Using Large DeviationTheory (LDT), we compute the probability error exponentsof a distributed scheme for detecting a correlated circularly-symmetric complex Gaussian process under the Neyman-Pearsonframework. Using an analog scheme, the sensors transmit scaledversions of their measurements several times through a multipleaccess channel (MAC) to reach the fusion center (FC), whose taskis to decide whether the process is present or not. In the analysis,we consider the energy constraint on each node transmission. Weshow that the proposed distributed scheme requires relatively fewMAC channel uses to achieve the centralized error exponentswhen detecting correlated Gaussian processes.