Exploiting Spatial Diversity for Anti-Jamming in LEO Mega-Constellations: Joint User Association and Coordinated Hybrid Beamforming

Low-Earth-orbit (LEO) mega-constellations are inherently vulnerable to uplink malicious jamming due to their open channel characteristics and predictable orbital trajectories. While beamforming and beamhopping can mitigate interference through spatial filtering and resource scheduling, existing approaches often decouple user association from physical-layer suppression, limiting their ability to exploit multi-satellite spatial diversity for uplink anti-jamming. To address this, we propose a unified cross-layer anti-jamming framework that jointly optimizes user-satellite association, user transmit coefficients, and satellite-side hybrid analog-digital combining to maximize the uplink weighted sum rate (WSR). Specifically, we formulate a mixed-integer non-convex optimization problem that captures the coupling between access-level association and physical-layer interference management. To tackle this computationally intractable problem, we develop a hierarchical optimization algorithm. In the inner loop, fractional programming and Riemannian manifold optimization are used to update the user transmit coefficients and hybrid receive combiners with monotonic convergence. In the outer loop, a Gale-Shapley based association strategy establishes a stable user-satellite matching under the fixed-preference assumption. Extensive simulations show that the proposed scheme outperforms competitive user association and hybrid beamforming baselines in large scale networks, approaches the ideal fully digital receiver under fixed user association, and nearly matches exhaustive user association search in small scale cases, providing empirical evidence of near optimal WSR performance.