Parallel Performance Analysis of Cyclic Correntropy for Energy-Efficient Wireless Communications
Keywords:Cognitive Radio, Cyclic Correntropy, Cyclostationary Process, Low Power, Spectrum Sensing, Parallel Scalability
Cognitive radio (CR) based systems appear as an efficient alternative for providing higher throughput in wireless systems, mainly through dynamic spectrum access (DSA), and automatic adaptation of the transmitter in terms of modulation format, coding rate, among other parameters, as a function of the communication channel conditions accessed by the system. In this context, processing techniques for spectrum sensing and automatic modulation classifications (AMC) has been widely developed in recent years. Among these techniques, cyclostationary correntropy can be applied to both problems the spectrum sensing and automatic modulation classification. However, this technique has a high computational cost that may increase significantly the system energy consumption. This work proposes a strategy to calculate the cyclic correntropy by using parallel processing on multi-core processors in order to decrease energy consumption in spectrum sensing tasks. A mathematical model is used to relate the power consumption of the processors with their parallel performance. The results demonstrate a decrease in the energy consumption of the proposed architecture.