<p>Biometric authentication schemes have gained popularity due to their potential for strong security measures and user accessibility. Despite its benefits, significant concerns remain about the security and privacy of biometric data. The proposed AEGIIS (Quantum-proof Biometric Authentication: A FramEwork UtilizinG BInary Lattices and HomomorphIc Encryption for Cancelable TemplateS) secures an individual’s biometric templates through efficient arithmetic operations on binary lattices and homomorphic encryption, utilizing random noise polynomials with uniform distribution. Three benchmark data sets, ORL, CASIA-Face V5 and deep funneled LFW, were used for security and performance evaluations, resulting in Error Equal Rates of 0.000714, 0.000438, and 0.006330, respectively. The generated cancelable templates demonstrate average entropy, NPCR, and UACI values of 7.9360, 99.62<InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="10791_2025_9631_Article_IEq1.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="15" /> </InlineMediaObject> <EquationSource Format="TEX">$\%$</EquationSource> <EquationSource Format="MATHML"><math> <mo>%</mo> </math></EquationSource> </InlineEquation>, and 33.59<InlineEquation ID="IEq2"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="10791_2025_9631_Article_IEq1.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="15" /> </InlineMediaObject> <EquationSource Format="TEX">$\%$</EquationSource> <EquationSource Format="MATHML"><math> <mo>%</mo> </math></EquationSource> </InlineEquation>, respectively, signifying substantially distorted templates that resist adversarial identification. The similarity score between freshly generated cancelable templates and stored CBT was evaluated against the predefined thresholds of Cosine Similarity (<InlineEquation ID="IEq3"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="10791_2025_9631_Article_IEq3.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="80" /> </InlineMediaObject> <EquationSource Format="TEX">$S_{\cos }\ge 0.95$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msub> <mi>S</mi> <mo>cos</mo> </msub> <mo>≥</mo> <mn>0.95</mn> </mrow> </math></EquationSource> </InlineEquation>), the Kolmogorov-Smirnov statistic Test (<InlineEquation ID="IEq4"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="10791_2025_9631_Article_IEq4.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="80" /> </InlineMediaObject> <EquationSource Format="TEX">$S_{\text {KS}}\le 0.05$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msub> <mi>S</mi> <mtext>KS</mtext> </msub> <mo>≤</mo> <mn>0.05</mn> </mrow> </math></EquationSource> </InlineEquation>) , and the Pearson Correlation Coefficient (<InlineEquation ID="IEq5"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="10791_2025_9631_Article_IEq5.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="84" /> </InlineMediaObject> <EquationSource Format="TEX">$S_{\text {corr}}\ge 0.85$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msub> <mi>S</mi> <mtext>corr</mtext> </msub> <mo>≥</mo> <mn>0.85</mn> </mrow> </math></EquationSource> </InlineEquation>), indicating successful authentication. The genuine and impostor distributions in the extensive data sets were used to validate the above authentication thresholds that were chosen through empirical testing. The proposed template protection method demonstrates a recognition accuracy of 99.5<InlineEquation ID="IEq6"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="10791_2025_9631_Article_IEq1.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="15" /> </InlineMediaObject> <EquationSource Format="TEX">$\%$</EquationSource> <EquationSource Format="MATHML"><math> <mo>%</mo> </math></EquationSource> </InlineEquation>. The AEGIIS scheme was designed to operate on the Ring-Learning-with-Errors hardness problem, which exhibits the super polynomial complexity of <InlineEquation ID="IEq7"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="10791_2025_9631_Article_IEq7.gif" Format="GIF" Height="19" Rendition="HTML" Resolution="72" Type="Linedraw" Width="74" /> </InlineMediaObject> <EquationSource Format="TEX">$\ {O(n \log n)}$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mspace width="4pt" /> <mrow> <mi>O</mi> <mo stretchy="false">(</mo> <mi>n</mi> <mo>log</mo> <mi>n</mi> <mo stretchy="false">)</mo> </mrow> </mrow> </math></EquationSource> </InlineEquation> and is resilient to both classical and quantum attacks and tested using indistinguishability analysis through IND-CPA analysis on ROM model, achieving post-quantum security with a key size of 256 bits. </p>

Quantum proof biometric authentication framework using binary lattices and homomorphic encryption for secure cancelable templates

期刊论文
出版模式：

开放获取
发表日期：

2025年6月27日

查看全文

S. Aarthi,
K. Geetha,
N. Sasikaladevi

摘要

Biometric authentication schemes have gained popularity due to their potential for strong security measures and user accessibility. Despite its benefits, significant concerns remain about the security and privacy of biometric data. The proposed AEGIIS (Quantum-proof Biometric Authentication: A FramEwork UtilizinG BInary Lattices and HomomorphIc Encryption for Cancelable TemplateS) secures an individual’s biometric templates through efficient arithmetic operations on binary lattices and homomorphic encryption, utilizing random noise polynomials with uniform distribution. Three benchmark data sets, ORL, CASIA-Face V5 and deep funneled LFW, were used for security and performance evaluations, resulting in Error Equal Rates of 0.000714, 0.000438, and 0.006330, respectively. The generated cancelable templates demonstrate average entropy, NPCR, and UACI values of 7.9360, 99.62 $\%$ $%$ , and 33.59 $\%$ $%$ , respectively, signifying substantially distorted templates that resist adversarial identification. The similarity score between freshly generated cancelable templates and stored CBT was evaluated against the predefined thresholds of Cosine Similarity ( $S_{\cos }\ge 0.95$ $S_{cos} \geq 0.95$ ), the Kolmogorov-Smirnov statistic Test ( $S_{\text {KS}}\le 0.05$ $S_{KS} \leq 0.05$ ) , and the Pearson Correlation Coefficient ( $S_{\text {corr}}\ge 0.85$ $S_{corr} \geq 0.85$ ), indicating successful authentication. The genuine and impostor distributions in the extensive data sets were used to validate the above authentication thresholds that were chosen through empirical testing. The proposed template protection method demonstrates a recognition accuracy of 99.5 $\%$ $%$ . The AEGIIS scheme was designed to operate on the Ring-Learning-with-Errors hardness problem, which exhibits the super polynomial complexity of $\ {O(n \log n)}$ $O (n log n)$ and is resilient to both classical and quantum attacks and tested using indistinguishability analysis through IND-CPA analysis on ROM model, achieving post-quantum security with a key size of 256 bits.