DSmT Applied to Seismic and Acoustic Sensor Fusion

In this paper, we explore the use of DSMT for seismic and acoustic sensor fusion. The seismic/acoustic data is noisy which leads to classification errors and conflicts in declarations. DSmT affords the redistribution of masses when there is a conflict. The goal of this paper
is to present an application and comparison on DSMT with other classifier methods to include the support vector machine(SVM) and Dempster-Shafer methods. The work is based on two key references (1) Marco Duarte with the initial SVM classifier application of the seismic and
acoustic sensor data and (2) Arnaud Martin in Vol. 3 with the Proportional Conflict Redistribution Rule 5/6 (PCR5/PCR6) developments. By using the developments of Duarte and Martin, we were able to explore the various aspects of DSMT in an unattended ground sensor scenario. Using the receiver operator curve (ROC), we compare the methods for individual classification as well
as a measure of overall classification using the area under the curve(AUC). Conclusions of the work show that the DSMT affords a lower false alarm rate because the conflict information is redistributed over the set masses and is comparable to other classifier results when using a
maximum decision forced choice.