A metadata deduplication framework associated with metadata placement and retrieval in edge-cloud using Adaptive Capsule Neural Network (A-CNet)
Year: 2026
Type: article
Abstract: With the rapid digitization and smart industrial systems, the volume of metadata stored in the edge-cloud has grown exponentially. Factors such as global connectivity and seamless access to information have further accelerated this growth. At the same time, many organizations collect and process vast amounts of data from multiple sources. Meanwhile, this growth leads to an increase in data redundancy, especially in edge-cloud storage that impacts data storage and analysis. For minimizing the duplicate metadata in edge-cloud, the proposed model employs a machine learning-based metadata deduplication technique. In this research study, a new metadata deduplication mechanism in the edge-cloud using machine learning is developed to eliminate duplicate metadata. At first, the metadata deduplication is done in the edge-cloud by the Adaptive Capsule Neural Network (A-CNet) model. The performance of this network is enhanced using the Improved Black-Winged Kite Algorithm (IBKA) that fine-tunes the relevant parameters. Another major challenge faced by geographically distributed edge-cloud storage systems is existing metadata placement methods often lead to high latency and increased security risks during data transmission and data access. To address this issue, the optimal metadata placement in the edge device is done by the developed IBKA. Before placing the metadata, the encryption process is performed using Elliptic Curve Cryptography (ECC) to protect the metadata from unauthorized access. The data location is identified, and the corresponding metadata is stored on an edge server associated with the node that is closest to the data’s virtual location in the spatial domain. The objective function, namely Euclidean distance between nodes is considered during metadata placement. Then, the data retrieval process is done by the Data Retrieval Protocol (DRP) approach, which helps in extracting the essential information based on the user query. Finally, the retrieved metadata is decrypted using the ECC algorithm. The overall performance of the developed method is evaluated with other techniques and encryption methods to validate its efficacy.
Source: Discover Computing
Language: English
Topic: Data Quality and Management
Field: Decision Sciences
Domain: Social Sciences
Open Access status: gold
Estimated APC paid: $1,520
OpenAlex ID: W7222220249