Radio access behavior (RAB) based cognitive radio classification and identification

Cognitive radio (CR) provides an open architecture for efficiently utilizing communication resources through flexible opportunistic access methods. However, such flexibility and dynamic access approach could lead to potential communication resource misuses and security threats. In order to successfully deploy a CR network and realize its benefits, distinguishing/classifying radio terminals and the radio behaviors is an important research issue. This paper explores unique radio characteristics in CR networks, radio access behavior (RAB) characteristics (radio access bandwidth, access time and access response time), in identifying CR terminals in a CR network. Using machine learning algorithms, the proposed RAB based CR classification method can be used for CR network monitoring and CR identification. A GNURadio/Universal Software Radio Peripheral (USRP) test bed is developed to implement and evaluate the performance of the RAB feature extraction and CR identification. The experimental results demonstrate that the proposed method is effective in CR classifications/identifications (differentiating radio types and radio terminals in a CR network).