TL;DR
This paper introduces a novel algorithm for simultaneously encrypting and watermarking vector maps, addressing the critical need for secure transmission and copyright protection of geospatial data. By combining a permutation-based encryption scheme with a coordinate-based watermarking approach, the method achieves robust protection against various attacks, particularly vertex manipulations, which are a weakness in existing solutions.
What Problem Does This Paper Address?
Vector maps are essential for numerous applications, from navigation to urban planning, but their increasing use in digital formats makes them vulnerable to leakage and unauthorized copying. While encryption can secure transmission and watermarking can protect copyright, a combined approach is needed to offer both functionalities efficiently. Existing Commutative Encryption and Watermarking (CEW) methods for vector maps often struggle with "vertex attacks" – operations like adding or deleting vertices – which can corrupt or remove embedded watermarks. This paper aims to bridge this gap by proposing a CEW algorithm that is specifically designed to be resilient to these types of attacks, while also maintaining strong security and copyright protection.
Methods
The proposed algorithm is built upon two core components:
- Permutation-Based Encryption Scheme: This scheme uses a "Double Random Position Permutation" (DRPP) strategy. Instead of directly altering vertex coordinates (which would interfere with watermarking), it rearranges the order of vertices. The encryption key is generated using a combination of the SHA-512 hash method and Gaussian random numbers for enhanced security. This method is designed to be robust against attacks that alter the spatial order of data.
- Coordinates-Based Watermarking Scheme: This scheme embeds watermark information directly into the normalized coordinates of the vector map's vertices. Before embedding, the watermark image is first shuffled using a 2D logistic chaotic map to improve its security. The embedding process utilizes a quantization method that allows each bit of the watermark to be mapped to at least one coordinate. This "multiple embedding strategy" is key to its robustness against vertex attacks, as the watermark can still be recovered even if some vertices are added or deleted.
The "commutative" aspect means that the order of encryption and watermarking does not affect the final outcome, allowing for watermark extraction before or after decryption.
Datasets
The experiments utilized shapefile data downloaded from OpenStreetMap (http://download.geofabrik.de/). These datasets represent real-world vector maps of varying sizes, regions, and object types (buildings, roads, waterways). A binary image of 32x64 pixels was used as the watermark. No new benchmarks were introduced; instead, existing datasets were used for evaluation.
Key Insights and Ideas
- Commutativity is Key: The ability to perform encryption and watermarking in any order without compromising the integrity of either operation is crucial for practical CEW systems. This paper achieves this by ensuring the encryption method doesn't alter vertex coordinates, allowing watermarking to operate on them.
- Vertex Attacks are a Major Weakness: Traditional CEW methods often fail when faced with modifications to the number of vertices in a vector map. The proposed algorithm's coordinate-based embedding with a "multiple embedding strategy" directly tackles this vulnerability.
- DRPP for Enhanced Encryption: The Double Random Position Permutation (DRPP) method, combined with secure key generation, provides a strong encryption layer that scrambles the spatial order of vector map data effectively.
- Chaotic Maps for Watermark Security: Using a 2D logistic chaotic map to shuffle the watermark image before embedding adds an extra layer of security, reducing pixel correlation and making the watermark harder to detect or tamper with.
Key Results
- Robustness to Vertex Attacks: The algorithm demonstrates high robustness against vertex addition (NC values remain above 0.8 even with 100% vertex addition) and vertex deletion (NC values above 0.8 with up to 60% deletion).
- Resistance to Geometric Attacks: The method shows good robustness against scaling and translation attacks, with NC values consistently at 1.0. However, it is not resistant to rotation attacks.
- Format Conversion and Reordering: The algorithm is robust to data format conversion (e.g., dwg, e00, gdb) and reordering of vector map data, with NC values of 1.0 in all tested cases.
- Commutativity Verified: Experiments showed that the Root-Mean-Square Error (RMSE) between E-Wed (Encrypted-Watermarked) and W-Eed (Watermarked-Encrypted) maps is 0, confirming commutativity. Similarly, Normalized Correlation (NC) values of 1 for extracted watermarks from both CEWed and D-Wed (Decrypted-Watermarked) maps demonstrate commutativity between decryption and extraction.
- Key Sensitivity: The algorithm exhibits strong key sensitivity, meaning even minor changes to the encryption key result in significantly different encrypted outputs and prevent correct decryption.
Limitations and Future Work
- Rotation Attack Vulnerability: The paper explicitly states that the proposed algorithm is not resistant to rotation attacks.
- Point Object Encryption: The encryption effect on point objects is noted as poor because permutation only changes storage order without impacting data structure.
- Future Work: The authors suggest that improving the CEW algorithm for point objects will be a focus of future research.
Summary updated