Using a probabilistic neural network and the nearest neighbor method to identify foreign ship radio stations

The paper presents the application of a probabilistic neural network and the nearest neighbor method for the identification of foreign ship radio stations. The research presented in the paper was conducted using real-world data.1. INTRODUCTION
The paper presents and compares two different solutions to the friend-or-foe identification problem: the nearest neighbor (NN) method and the probabilistic neural network (PNN). Both solutions are classic memory-based object identification methods. The construction of an identification system in both cases involves storing a certain number of training images representing all identified objects in memory. However, the friend-or-foe identification task assumes a lack of training images for objects considered foreign. In this case, the identification system must be based solely on data about friendly objects. To check how both methods cope in a situation where information about one of the identified classes is missing, experiments were conducted during which the task of both methods was to identify ship radio stations.
2. METHODS USED DURING THE RESEARCH
2.1 Nearest neighbor method
The simplest way to identify an object is to compare its signature with the signatures of objects stored in the system. This can be done by searching the system database for the signature most similar to the signature of the object being identified. (…)

Using a probabilistic neural network and the nearest neighbor method to identify foreign ship radio stations – PDF file for download