In this paper, we propose a promising signal-sensing method based on both single- and multiple-antenna covariance matrices of the received signal. In comparison with microwave diagnostics or energy detection, the proposed method does not need prior information of signal and noise. Therefore, it can be widely used in noise power uncertainty scenarios, such as signal transmission in hypersonic plasma sheath, with symmetry to the signal-sensing method in cognitive radio. Designing an efficient signal-sensing method that supports both the application for the field of hypersonic plasma sheath communication or cognitive radio is still an attractive subject deserving of study in symmetry journals. Theoretical analysis and the implementation algorithm of a single-antenna covariance matrix are carried out for the proposed method, with a possible extension to multiple-antenna situations. Simulations show the proposed method has excellent performance, even for very low signal-noise-ratio (SNR) signals, and is competitive with state-of-the-art algorithms.