In this paper an advanced version of the extensive cancellation algorithm (ECA) is proposed for robust disturbance cancellation and target detection in passive radar. Firstly some specific limitations of previous ECA versions are identified when dealing with a highly time-varying disturbance scenario in the presence of slowly moving targets. Specifically, the need to rapidly adapt the filter coefficients is shown to yield undesired effects on low Doppler target echoes, along with the expected partial cancellation. Therefore a sliding version of the ECA is presented which operates on partially overlapped signal batches. The proposed modification to the original ECA is shown to appropriately counteract the limitations above by taking advantage of a smooth estimate of the filter coefficients. An efficient implementation is also discussed to limit the corresponding computational load. The benefits of the proposed approach are demonstrated against real data sets accounting for quite different passive radar applications.