Abstract: ‍ ‍Compared to traditional radars， phased array radars can generate multiple beams simultaneously， flexibly change beam direction， and have been widely used for multi-target tracking. To support the task requirements of subsequent nodes intercepting and striking enemy targets in large-scale cluster target collaborative detection scenarios， phased array radars need to track higher-priority targets in the airspace to ensure faster fire control accuracy within a specified time. However， radar detection resources are limited when too many targets are in the airspace， making it difficult to complete the specified tracking task. Therefore， this study proposes a resource allocation algorithm for phased array radars under resource-constrained conditions to address this issue. First， we derived the Predicted Conditional Cramer-Rao Lower Bound （PC-CRLB）， which included target allocation and power optimization， and used it as a metric for tracking accuracy. Subsequently， we considered the tracking capacity and accuracy， with the optimization objectives of maximizing the number of targets that meet the specified tracking accuracy and minimizing the weighted average tracking error of multiple targets. In addition to phased array radar system resources， a joint optimization model for target allocation and power under large-scale target tracking was established， and adaptive joint optimization configuration was performed for target allocation variables and transmission power variables. To solve this optimization problem， we employed a two-step decomposition method， breaking it down into target allocation and power optimization subproblems. We also used activation functions to smoothly approximate the non-smooth and non-convex objective functions. Then， we solved the problem using the spectral projected gradient method. Finally， the simulation experiments demonstrated that the proposed algorithm outperformed the compared algorithms in tracking more targets to the specified accuracy within a given time in various scenarios.