Wireless Traffic Reasoning Method Based on Support Vector Machine according to the Probability of Incumbent User’s Channel Occupancy
Han-Sol Kim, Deok-Won Yun, Joo-Pyoung Choi, Dong-Hwan Yoon, Jung-Kyu Sun, Ho-Suk Jeon, Won Cheol Lee · The Journal of Korean Institute of Electromagnetic Engineering and Science · 2019
In this paper, we propose a case-based reasoning Cognitive Radio(CR) engine that uses limited resources efficiently in military tactical wireless communication environments. A CR engine should be able to learn and infer, and thus predict the available channel information of a secondary user based on information about traffic usage. To be able to do so, a low probability of channel collision should be associated with the engine. The engine should, thereby, be able to indicate the probability of channel occupancy of an incumbent user who requires interference protection. We used a Support Vector Machine(SVM) to measure the histogram-type wireless traffic usage environment in accordance with the change in the probably of channel occupancy of the incumbent user. SVM is a sorting algorithm of machine learning. Next, we calculated the histogram's skewness and kurtosis for traffic modeling. Finally, to analyze the performance of the SVM-based wireless traffic usage environment, we compared the inference accuracy and time complexity of the proposed SVM with those of k-Nearest Neighbors(k-NN).