Abstract: Given the increasing proliferation of drone systems， monitoring low-altitude， slow， small （LSS） targets in urban environments is an increasingly urgent requirement. However， in complex urban settings， factors such as occlusion by buildings and multipath propagation prevent passive radar systems from characterizing prior information on measurement errors accurately. This leads to a degradation in tracking precision for maneuvering targets and can even cause filter divergence. To address these challenges， we propose a joint tracking strategy that integrates an improved adaptive extended Kalman filter with the interacting multiple model （IMM） algorithm. First， numerical stabilization is applied to ensure the positive definiteness of the immediately estimated measurement covariance matrix. Second， a multidimensional vector is constructed as a control factor by comparing the theoretical estimates and the actual computed values of the filter innovation covariance， which enables dynamic perception of changes in measurement noise statistics. Suitable weighting coefficients are then assigned to each dimension for refined online correction of the measurement covariance matrix. Finally， the entire adaptive filtering module is deeply integrated with the IMM framework. This allows the algorithm to handle the dual uncertainties that arise from target maneuvers and environmental variations. The results of a simulation indicate that the proposed SHAEKF-IMM algorithm was able to capture and respond to nonstationary changes in measurement noise covariance more promptly and accurately compared to conventional algorithms while tracking maneuvering targets using radars with an external radiation source. The proposed approach significantly suppressed filtering errors caused by model mismatch through an effective online estimation and compensation mechanism. Thus， it demonstrated superior robustness and tracking accuracy across different maneuvering scenarios.