Design of a Blockchain-Powered Biometric Template Security Framework Using Augmented Sharding

Introduction

Securing biometric signatures is a multidomain task, which involves cryptanalysis, privacy protection, traceability integration, signal processing, classification, etc. In order to design a high efficiency secure biometric signature processing system, a wide number of system capabilities are needed. These includes but are not limited to, data security, resilience against tampering, high speed traceability, low computational complexity, and reduced power requirements. An example of such a model is depicted in figure 1, wherein processes like template enrolment, identity claim, and exception handling can be observed. The model utilizes user input for capturing biometric information including, fingerprints, facial data, etc. and provides this information for quality assessment, and feature extraction. The extracted features are given to a template database for storage, which is connected to a system administrator for managerial purposes. The connection between system admin and template matching module is facilitated using an application programming interface (API) layer. This layer exposes the data to various 3^rd parties, which might be prone to spying, spoofing, or tampering attacks. Thus, there is a need for a security layer to be integrated between the system and any incoming and outgoing connections from it. Thus, the user data capturing connection, system administrator connection, application connection, and any other connection(s) must be secured using a high-performance security layer, which provides high-speed data read/write capabilities.

In order to design such a security layer, various cryptographic, key-exchange, privacy preservation, and machine learning approaches are proposed by the researchers over the years. A survey of these approaches can be observed from the next section of this text, wherein various nuances, advantages, limitations, and future issues are discussed.

Figure 1.

An identity verification model using biometrics

Based on this discussion, it was observed during blockchain mining, each block needs to be scanned for evaluating unique and rule-based hash values. This requires substantial delay with increasing chain length, which reduces scalability of the system. To improve scalability, machine learning models are used, which aims at reducing redundant calculations during storage and retrieval, thereby increasing storage and retrieval speeds. Some of these models have limited integration capabilities with blockchain, due to their internal build structure. Other interfaceable models do not provide a significant performance improvement because large length blockchains require mandatory hash verification and rule validations. Thus, problem statement of this text is to maintain high security with good scalability under different network conditions. To achieve a solution for this problem statement, section 3 discusses design of the proposed blockchain powered biometric template security framework using augmented sharding (BLTSAS), and its performance evaluation. This evaluation also includes comparison with various state-of-the-art methods, which assists in estimating any performance gaps in the proposed model. Finally, this text concludes with some interesting observations about the proposed work, and recommends methods to further improve its performance.