Low-Power Wide Area Networks (LPWANs) represent a category of wireless technologies hailed for their efficiency in facilitating communication for Internet of Things (IoT) applications. This efficacy is attributed to their characteristics of low power consumption, extensive wireless transmission range, and cost-effectiveness. Despite these notable advantages, LPWANs exhibit drawbacks such as limited processing power, modest transmission rates, and notably constrained payload sizes, posing challenges for encryption techniques. The inadequacy of existing cipher-chaining encryption methods for LPWANs is underscored by their dependency on high computing power and payload capacity. To address this issue, this paper introduces an innovative chaining encryption approach tailored for LPWAN IoT technology. The proposed method incorporates a key synchronization mechanism based on deep learning algorithms. The effectiveness of this approach has been rigorously assessed through case studies and experiments. Results from the experiments reveal a commendable performance of the proposed approach, achieving an average accuracy of 99% in key resynchronization following message loss. These outcomes establish the proposed method as a highly competitive solution within the realm of encryption approaches for LPWANs.