2020/03/05 by Juan Guillermo Tamayo Maya, Maya, Juan, Leonardo Rey Vega +1
Computer Science · Decision Sciences · #Advanced Statistical Process Monitoring #Distributed Sensor Networks and Detection Algorithms #FOS: Electrical engineering #Signal Processing (eess.SP) #Target Tracking and Data Fusion in Sensor Networks #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2003.02490
openalex publication_date 2020/03/05 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
We consider a distributed detection problem where measurements at each sensor\nfollow a general parametric distribution. The network does not have a central\nprocessing unit or fusion center (FC). Thus, each node takes some measurements,\ndoes some processing, exchanges messages with its neighbors and finally makes a\ndecision (typically the same for all nodes) about the phenomenon of interest.\nThe problem can be formulated as a composite hypothesis test with unknown\nparameters where, in general, a uniformly most powerful test does not exist.\nThis leads naturally to the use of the Generalized Likelihood Ratio (GLR) test.\nAs the measurements follow a general parametric distribution (which could model\nspatial dependence of the data), the implementation of fully-distributed\ndetection procedures could be demanding in network resources. For this reason,\nwe study the use of a simpler test (referred as L-MP) which uses the product of\nthe marginals of the measurements taken at each node, where the unknown\nparameters are easily estimated with only local measurements. Although this\nsimple proposal still requires network-wide cooperation between nodes, the\nnumber of communications is significantly reduced with respect to the GLR test,\nmaking it a suitable choice in severely resource-constrained sensor networks.\nThis simpler test does not exploit the full parametric model of data, so, it\nbecomes important to analyze its statistical properties and its potential\nperformance loss. This is done through the analysis of the L-MP asymptotic\ndistribution. Interestingly, despite the fact that the L-MP is simpler and more\nefficient to implement than the GLR test, we obtain some conditions under which\nthe L-MP has superior asymptotic performance to the GLR test. Finally, we\npresent numerical results for a fully-distributed spectrum sensing application\nfor cognitive radios.\n