Welch FFT Segment Size Selection Method for Spectrum Awareness System

Abstract

We create a practical method to set the segment size of the Welch FFT for wideband and long-term spectrum usage measurements in the context of hierarchical dynamic spectrum access (DSA). An energy detector (ED) based on the Welch FFT can be used to detect the presence or absence of primary user (PU) signal and to estimate the duty cycle (DC). In signal detection with the Welch FFT, segment size is an important design parameter since it determines both the detection performance and the frequency resolution. Between these two metrics, there is a trade-off relationship which can be controlled by adjusting the segment size. To cope with this trade-off relationship, we define an optimum and, more easy to analyze sub-optimum segment size design criterion. An analysis of the sub-optimum segment size criterion reveals that the resulting segment size depends on the signal-to-noise ratio (SNR) and the DC. Since in practice both SNR and DC are unknown, proper segment setting is difficult. To overcome this problem, we propose an adaptive segment size selection (ASSS) method that uses noise floor estimation outputs. The proposed method does not require any prior knowledge on the SNR or the DC. Simulation results confirm that the proposed ASSS method matches the performance achieved with the optimum design criterion.

Smart Spectrum for Future Wireless World

“…Therefore, in fact it is difficult to set the appropriate segment size. To resolve this issue, a method to set the appropriate segment size based on the noise floor (NF) estimation was proposed in [37]. In this method, NF estimation is used, and it has been shown that the appropriate segment size can be achieved without SNR information.…”

Welch FFT Segment Size Selection Method for FFT Based Wide Band Spectrum Measurement

“…Specifically, ED does not need any specific information about observed signal, such as modulation scheme. Welch FFT-based ED is an effective method for achieving proper spectrum usage detection [15], [16]. Welch FFT consists of three steps: segmentation of data sequence with a specific segment size, calculation of multiple power spectra and averaging of the power spectra.…”

“…Therefore, proper segment size for Welch FFT is important as there is a trade-off between the detection performance and frequency resolution in terms of segment size. Accurate spectrum usage detection and high frequency resolution are required for accurate DC estimation and accurate recognition of bandwidth of spectrum occupied by PUs, i.e., accurate recognition of the vacant spectrum, which is usually denoted by white space (WS), respectively [16].…”

“…For the issue in terms of segment size, we formulated the optimum segment size in terms of a criterion which is determined by white space detection ratio (WSDR) performance with a constraint determined by sensitivity required for spectrum usage detection [16]. WSDR is defined by the ratio of estimated WS and actual WS in time and frequency domains, therefore WSDR = 1 indicates ideal situation, WSDR > 1 indicates overestimation of WS, and WSDR < 1 indicates underestimation of WS.…”

“…The sensitivity in the constraint is quantified by root mean square error (RMSE) of DC estimation. In fact, the optimum segment size depends on signal-to-noise ratio (SNR) which makes practical and optimum segment size setting difficult [16].…”