Device resolution is essential in cybersecurity, fraud detection, and medical technology, enabling precise device identification and authentication. However, most existing device fingerprinting methods are fragile, exhibiting high sensitivity to changes in device settings or configurations, which disrupts reliable device matching and hinders device-sharing detection. To our knowledge, current methods do not integrate link analysis of account-device relationships, limiting their consistency and effectiveness in device resolution. This paper introduces a Novel Adaptive Device Resolution Algorithm (NADER), which leverages link analysis of account-device relationships and employs similarity measures to enhance device matching accuracy. Due to the lack of a labeled dataset for this task, we constructed a dataset by simulating dynamic device environments. Empirical results show that NADER achieves 22% higher true positive matches compared to baseline device resolution methods. Additionally, NADER’s privacy-preserving, adaptive design supports robust performance in dynamic device environments, providing a more resilient and configurable solution.