<p>In recent years, the rapid increase in the number of devices connected to the internet has created a major challenge for asset management. It assesses whether the cyber-attacks targeting IoT devices, including items like pinhole cameras, compromise privacy. Traffic-based identification of IoT device types has been shown to be an effective method for asset management and safeguarding against cyber-attacks. However, given the rapid expansion of IoT devices, accurately identifying newer, innovative devices remains a challenge, even with the current mainstream identification methods. However, due to the limited traffic generated by these devices, most identification algorithms struggle to accurately detect a significant proportion of lightweight IoT devices. In order to address these issues, the data are collected from the IoT devices in the dataset. The pre-processing, min-max normalization, Local Outlier Factor, and Multiple Imputation using Chained Equations are used to normalize the data and replace the missing values. Recursive Feature Elimination (RFE) is used to select the most relevant features from the pre-processed data, and Walrus optimization is employed in the feature selection process. Finally, a Bidirectional Gated Recurrent Unit (BiGRU) is used to train the best features from the feature selection process in order to accurately detect IoT devices based on trajectory data. The proposed method achieves a performance accuracy of 97.40%. It is a better choice for IoT device identification based on trajectory data when compared to existing systems.</p>