Wideband Cyclostationary Spectrum Sensing: An Adaptive Algorithm-Based Approach

In this paper a cyclostationary​-​based wideband spectrum sensing method is proposed. The received signal passes through a rough and flexible filter with its effective band tuned to a specific part of the received signal spectrum. This part belongs to a target signal which potentially exists in the received signal. Some cyclic frequencies of the target signal are employed to derive a normalized least mean square (NLMS) adaptive algorithm that estimates the output of the filter from its frequency shifted samples. If the target is absent in the received signal, the norm of the weights of the NLMS algorithm is almost zero. On the other hand, in the case of presence of the target, the norm of the weights will be greater than a certain threshold. The procedure is repeated to cover the entire band of the received signal and therefore it detects all cyclostationary signals with known cyclic frequencies in the received signal. The overall system is very easy to implement and fast and its performance is comparable to other spectrum sensing counterparts.