Abstract: In passive radar target detection, conventional sparse model-based methods typically require precise estimation of direct-path signals, thereby leading to increased system complexity and elevated computational burden. Furthermore, they frequently fail to address the issue of target masking by strong clutter. To overcome these limitations, a novel target detection method is proposed for orthogonal frequency division multiplexing-based passive radar in clutter-heavy environments. This method not only eliminates the need for precisely estimating the direct-path wave signal as a reference signal but also enables the detection of targets masked by strong clutter. The prior information of the method is comprised solely of deterministic and known pilot information. And the strong zero-frequency clutter in the real received surveillance signal can be suppressed using only the pilot information. The new clutter-suppressed surveillance signal is correlated with a newly formulated reference signal, also derived solely from the pilot information, to perform range compression, which enables the establishment of a time-invariant sparse model for range-spectrum domain data. Subsequently, a sparse optimization problem is constructed and solved to obtain the range-doppler spectrum where target information is prominently revealed. Simulation and field experiment results demonstrate that the proposed method achieves reduced computational load and superior detection performance compared to conventional sparse models and existing time-invariant sparse counterparts.