A Lossless Data-Hiding based IoT Data Authenticity Model in Edge-AI for Connected Living

Edge computing is an emerging technology for the acquisition of Internet-of-Things (IoT) data and provisioning different services in connected living. Artificial Intelligence (AI) powered edge devices (edge-AI) facilitate intelligent IoT data acquisition and services through data analytics. However, data in edge networks are prone to several security threats such as external and internal attacks and transmission errors. Attackers can inject false data during data acquisition or modify stored data in the edge data storage to hamper data analytics. Therefore, an edge-AI device must verify the authenticity of IoT data before using them in data analytics. This article presents an IoT data authenticity model in edge-AI for a connected living using data hiding techniques. Our proposed data authenticity model securely hides the data source’s identification number within IoT data before sending it to edge devices. Edge-AI devices extract hidden information for verifying data authenticity. Existing data hiding approaches for biosignal cannot reconstruct original IoT data after extracting the hidden message from it (i.e., lossy) and are not usable for IoT data authenticity. We propose the first lossless IoT data hiding technique in this article based on error-correcting codes (ECCs). We conduct several experiments to demonstrate the performance of our proposed method. Experimental results establish the lossless property of the proposed approach while maintaining other data hiding properties.

DIGDH: A Novel Framework of Difference Image Grafting Deep Hiding for Image Data Hiding

Data hiding is the technique of embedding data into video or audio media. With the development of deep neural networks (DNN), the quality of images generated by novel data hiding methods based on DNN is getting better. However, there is still room for the similarity between the original images and the images generated by the DNN models which were trained based on the existing hiding frameworks to improve, and it is hard for the receiver to distinguish whether the container image is from the real sender. We propose a framework by introducing a key_img for using the over-fitting characteristic of DNN and combined with difference image grafting symmetrically, named difference image grafting deep hiding (DIGDH). The key_img can be used to identify whether the container image is from the real sender easily. The experimental results show that without changing the structures of networks, the models trained based on the proposed framework can generate images with higher similarity to original cover and secret images. According to the analysis results of the steganalysis tool named StegExpose, the container images generated by the hiding model trained based on the proposed framework is closer to the random distribution.

Fuzzy Proximity based Robust Data Hiding Scheme with Interval Threshold

In 2020, Ashraf et al. proposed an interval type-2 fuzzy logic based block similarity calculation using color proximity relations of neighboring pixels in a steganographic scheme. Their method works well for detecting similarity, but it has drawbacks in terms of visual quality, imperceptibility, security, and robustness. Using Mamdani fuzzy logic to identify color proximity at the block level, as well as a shared secret key and post-processing system, this paper attempts to develop a robust data hiding scheme with similarity measure to ensure good visual quality, robustness, imperceptibility, and enhance the security. Further, the block color proximity is graded using an interval threshold. Accordingly, data embedding is processed in the sequence generated by the shared secret keys. In order to increase the quality and accuracy of the recovered secret message, the tampering coincidence problem is solved through a post-processing approach. The experimental analysis, steganalysis and comparisons clearly illustrate the effectiveness of the proposed scheme in terms of visual quality, structural similarity, recoverability and robustness.