Implementation Issues of Adaptive Energy Detection in Heterogeneous Wireless Networks.

Spectrum sensing (SS) enables the coexistence of non-coordinated heterogeneous wireless systems operating in the same band. Due to its computational simplicity, energy detection (ED) technique has been widespread employed in SS applications; nonetheless, the conventional ED may be unreliable under environmental impairments, justifying the use of ED-based variants. Assessing ED algorithms from theoretical and simulation viewpoints relies on several assumptions and simplifications which, eventually, lead to conclusions that do not necessarily meet the requirements imposed by real propagation environments. This work addresses those problems by dealing with practical implementation issues of adaptive least mean square (LMS)-based ED algorithms. The paper proposes a new adaptive ED algorithm that uses a variable step-size guaranteeing the LMS convergence in time-varying environments. Several implementation guidelines are provided and, additionally, an empirical assessment and validation with a software defined radio-based hardware is carried out. Experimental results show good performance in terms of probabilities of detection ( P d > 0 . 9 ) and false alarm ( P f ∼ 0 . 05 ) in a range of low signal-to-noise ratios around [ - 4 , 1 ] dB, in both single-node and cooperative modes. The proposed sensing methodology enables a seamless monitoring of the radio electromagnetic spectrum in order to provide band occupancy information for an efficient usage among several wireless communications systems.