Industrial wireless networks (IWNs) are characteristically different from traditional wireless systems due to harsh radio frequency (RF) environments and applications that impose high real-time and reliability constraints. One of the promising technologies for enabling IWNs is the cognitive radio. This paper designs a novel subcarrier modulation-based cooperative spectrum sensing (SMCSS) scheme for cognitive industrial wireless networks (CIWNs). Based on the orthogonal frequency division multiplexing physical layer, the local decisions are sent to the fusion center orthogonally in the frequency domain via physical layer signaling with binary amplitude shift keying symbols, which significantly reduces the reporting delay of the SMCSS. Furthermore, a two-level decision fusion method is proposed to cope with the harsh control channel that suffers from strong Gaussian noise and Rayleigh fading. Sensing parameters of the two-level decision fusion are optimized for minimizing the total sensing error rate of CIWNs. Finally, a reputation updating mechanism is combined with the SMCSS to improve its robustness against unreliable subcarriers due to RF hostile industrial environments.INDEX TERMS Industrial wireless networks, cognitive radio, spectrum sensing, decision fusion, total sensing error rate.