Abstract
dc:description.abstract<p>Evaluation of Distributed Sensor Networks (DSN's) for optimal detection using Bayesian cost formulation methods has been the objective of several studies. There have been a few studies on DSN evaluation using an information theoretic approach, wherein an assymetric channel models a detector. We look at the multi-sensor system at the receiver as a black-box and model it as an assymetric channel whose cross over probabilities depend on the probabilties of detection and of false alarm. These probabilities in turn depend on the thresholds of the local detectors and on the fusion rule used. We consider the case wherein the receiver has control over the value of the probability of the signal being present, by influencing trans?mitter coding. The probability to be used at the receiver is the one that solves the MED (Minimum Equivocation Detetion) problem. Minimizing the equivocation between the input and output is the same as maximizing the mutual information. The input probability, P0, at the receiver that maximizes the mutual information is then achieved by having an encoder between the transmitter and the multi-sensor system at the receiver. The only factor in encoding is the proportion of 0's to 1's. Variable length codes with different ordering of 0's and 1's are possible to get the same performance. Results show that as we move towards an optimum ROC (Receiver Operating Characteristic) curve, the mutual information attains a higher value. The value of P0 that helps attain this value is then derived by encoding the transmitter output.</p>
Degree
thesis:*- Name thesis:degree_name
- Master of Science in Electrical Engineering - (M.S.)
- Discipline thesis:degree_discipline
- Electrical and Computer Engineering
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Mehrotra, Vivek
- Contributors dc:contributor
-
- Nirwan Ansari
- Joseph Frank
- Edwin Hou
Subjects
dc:subject × 4Identifiers
dc:identifier.*- Repository record dc:identifier
- https://digitalcommons.njit.edu/theses/2326