Performance Optimisation of Soft and Hard Spectrum Sensing Schemes in Cognitive Radio

Cognitive radios represent a promising technique to address the problem of efficient spectrum utilisation. However, their performance is limited by their capability to sense the spectrum. This letter proposes a fair comparison between soft and hard spectrum sensing schemes, based on the practical assumption of limited time resources, which was never explored in previous studies. By considering the achieved detection probability as the performance metric and the false alarm probability as constraint, this work analytically proves that the hard scheme can represent a favourable solution, especially in short sensing times and/or large numbers of users. The sufficient condition by which the soft scheme outperforms the hard scheme is also introduced. This condition is represented by the number of collected samples from the channel in a closed-form expression, which depends on the sensing time, the Signal-to-Noise-Ratio (SNR), the number of available users and the predefined false alarm probability threshold. All these factors are independently evaluated throughout this work. The analysis is carried out through mathematical equations and confirmed by simulation results.