In cognitive positioning systems, spectrum sensing methods play an important role to understand the surrounding spectrum. Due to their good performance under noisy environments, cyclostationary methods are commonly used to characterise the received signals. These methods require a higher computational cost and high sampling rates [1]. With that in mind, this paper uses real measurement data, acquired in an office environment, at different sampling rates, including rates below the Nyquist rate. The motivation is to show that the implementation burden of these methods can be reduced by using lower sampling frequencies, since the cyclic properties of the signals are still visible.