Proposal and hardware implementation of smart threshold setting methods for spectrum sensing

For spectrum sensing in very low SNR regime, noise uncertainty (Nu), either arises from change of environment or hardware noise of a sensing device, could make proper setting of threshold difficult and therefore, is recognized as a challenge for achieving robust sensing. This paper proposes two smart threshold setting methods, called dynamic threshold selection (DTS) and realtime threshold setting (RTTS), respectively, to overcome the NU produced by hardware system noise in spectrum sensing. In DTS based sensing, a threshold table, which shows corresponding thresholds for non-overlapping noise variance ranges, is produced by the sensing device in advance through a study process; during sensing, the sensing device estimates the instantaneous noise variance and by mapping which to the threshold table it finds the proper threshold. In RTTS based sensing, the sensing device isolates itself from receiving any signals, computes threshold statistics immediately prior to sensing and accordingly decides the proper threshold for subsequent sensing. Computer simulation results show that by adopting the proposed methods, degradation of sensing performance caused by Nu can be largely compensated. The hardware implementation of DTS into a sensing prototype with complete hardware and software design is also shown, and the hardware test results further verify the novelty of proposed methods.