A Blind OFDM Signal Detection Method Based on Cyclostationarity Analysis

In this paper, we present a novel technique to sense, blindly infer signal features [FFT size, cyclic prefix (CP) length], and detect OFDM signals based on second order cyclostationarity analysis. First, we infer accurate FFT size and CP length from the sensed signal based on cross correlation, through considering FFTs of different size (2 ^L ) and CPs length. In our experimental study, we assume that CP length in the sensed OFDM signal could be 5–15 % of the FFT size {64, 128, 256, 512, 1024, 2048 and 4096} used at primary user level. We successfully estimate accurate FFT size and CP length, and carry out performance analysis of the proposed approach at various channel conditions, and the effect of increase in sample length (frames) of the sensed signal. Additionally, we derive a recursive procedure to calculate the cross-correlation at sample (l + 1), using the cross-correlation value at sample (l) and a few mathematical operations. We have also tested MAX values distribution for FFT size and CP, whether inferred parameters are valid or not, by finding the confidence of estimation. With experimental results we evaluated that the proposed approach can successfully measure unknown OFDM signal parameters and detect OFDM signals.