Conventional active guidance laws can achieve asymptotic or finite-time convergence of the line-of-sight (LOS) angle error. However, the convergence time is relatively conservative. In this article, we propose predefined-time (PT) sliding-mode (SM) convergence active defense guidance laws. Firstly, the proposed guidance law is designed based on a nonlinear LOS guidance framework, eliminating the assumption of small flight-path angles. Then, robust PT control theory is employed to achieve PT convergence of the SM surface. Moreover, two levels of cooperation between the target and the defender are introduced to enhance guidance performance. To minimize maximum overload, static optimization is incorporated in two-way cooperation by a performance index. The obtained guidance law does not require an observer or time-to-go estimation as compared to previous work. Additionally, our method can be applied not only in planar plane but also extended directly to three-dimensional geometry. Finally, we validate the effectiveness of our proposed strategies through comparative simulations.