A cross-validation based semi-analytical performance evaluation for high order modulations

Higher order quadrature amplitude modulation (HOQAM) is one of the key technologies used in advanced mobile communication systems to achieve high data rates. However, in such systems, performance evaluation can be very complicated and time consuming due to the increase of constellation size. As a solution, a QAM scheme is seen as a combination of two pulse amplitude modulations (PAM). Nevertheless, the system remains complex especially with very high constellation sizes. Therefore, Monte-Carlo simulation method is impractical to obtain the expected system performance. In this paper, we propose a semi-analytical performance evaluation methodology for a PAM modulation. Our approach is based on the probability density estimation using kernel technique in joint application with cross-validation (CV) criterion. We derive both semi-analytical symbol error probability (SEP) and semi-analytical bit error probability (BEP) for several modulation schemes. Simulation results show that, in terms of SEP and BEP metrics, our technique is accurate compared with theoretical error probabilities.