ABSTRACT The distributed modulated wideband converter (DMWC) is a kind of sub-Nyquist sampling system which randomly moves the sparse wideband signal spectrum to the baseband to realise low-speed sampling. DMWC regards each sensor node as a sampling channel and combines compressed sensing technology with a broadband cooperative spectrum sensing network. However, DMWC contains phase and transmission attenuation, which results in joint sparsity destroyed. Therefore, the original signal cannot be reliably recovered. In this paper, we combine measurement matrix with a phase shift to develop a phase shift calibration algorithm (PSCA) based on DMWC, a calibration matrix that is constructed to eliminate the influence of phase shift on recovery rate. Firstly, random waveforms of different channels in the DMWC system correspond to the row vectors of the measurement matrix. We apply the frequency-domain method to construct a calibration matrix. Secondly, we estimate the phase shift in each channel and write it as a complex diagonal matrix. Finally, we consider balancing the phase shift in each channel, the phase shift of each channel multiplied by the corresponding row of the measurement matrix. Theoretical analysis and numerical simulations show that the algorithm improves the recovery rate to 98% or higher in lower SNR.