Decision-Making Method for Communication Parameter Selection via Support Vector Regression

Cognitive radio (CR) is recognized as the development direction of radio system in the future. Current researches are mainly focused on spectrum sensing and dynamic spectrum access. However, intelligent decision-making should be the core feature of CR. The design goal of the decision-making engine is to determine the communication waveform configuration (including carrier frequency, modulation mode, coding rate, etc.) which is suitable for the current radio environment. This paper introduces the theoretical basis of the support vector regression (SVR) algorithm. Then, a waveform parameter decision-making method for the orthogonal frequency division multiplexing (OFDM) communication system is proposed based on it. Using this method, we can choose waveform parameters which optimize the communication performance under any channel conditions. The simulation results illustrate the decision-making model can give reasonable decision surfaces in various wireless channels and then obtain decision results.