Local Biometric Authentication via Hyper-Sphere Vector Projection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In a post-COVID work environment, employers face challenges in authenticating workers' health status and identity without compromising privacy, as central databases are not maintained due to privacy concerns, and malicious validators may attempt to log valid credentials to bypass the system.
Innovation Solution
A system that uses non-deterministic biometric data, such as voice and face images, to generate authentication credentials locally on a mobile device, without relying on a central database, employing machine learning models to create characteristic identity vectors and authenticate users through scannable codes with embedded timestamps and location information to prevent replay attacks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a central database is used to store workers' health and identification information, then authentication can be performed efficiently, but privacy issues arise and security risks increase
Solution Approach 1:
The patent extracts the biometric template generation and storage functionality from a centralized database architecture and relocates it to distributed edge devices (mobile phones, tablets, wearables). Each user's biometric template is stored locally on their own device or on devices they control, eliminating the need for a centralized database that stores sensitive biometric information. This extraction resolves the contradiction by maintaining authentication efficiency through local processing while eliminating privacy and security risks associated with centralized storage.
Solution Approach 2:
The patent segments the authentication system into multiple independent components: biometric data collection on the user device, local template generation using neural networks, and distributed verification across multiple edge devices. This segmentation allows authentication to be performed efficiently through local operations while distributing security risks, as no single point of failure or privacy vulnerability exists in the segmented architecture.
2Speed
If biometric data is stored centrally for authentication, then verification speed is improved, but the system becomes vulnerable to malicious logging and credential theft
Solution Approach 1:
The patent implements self-service authentication where the user's own device generates and stores their biometric template, and performs local verification operations. The device autonomously compares incoming biometric data against its stored template without requiring external validation. This self-service approach maintains verification speed through local processing while eliminating security vulnerabilities, as the authentication credential cannot be stolen or logged by external validators when the verifying validator is the user's own device.
Solution Approach 2:
The patent inverts the traditional authentication model by making the user's own device the verifying validator rather than an external server. Instead of the system verifying the user, the user's device verifies itself using locally stored biometric templates. This inversion resolves the security contradiction by ensuring that no external entity can intercept or log authentication credentials, as the verification process occurs entirely within the user's controlled environment.
3Reliability
If a validator system is implemented to authenticate workers, then access control is improved, but the risk of malicious validators logging credentials increases
Solution Approach 1:
The patent extracts the validator functionality from external systems and embeds it within the user's own device. The biometric template generation and verification operations are taken out of external validator systems and performed locally on the user's mobile device or tablet. This extraction eliminates the security vulnerability of external validators logging credentials, as the authentication process occurs entirely within the user's controlled environment without external intervention.
Solution Approach 2:
The patent introduces cryptographic intermediaries (digital signatures, hash functions, and encrypted communications) between the biometric processing and access control systems. These cryptographic intermediaries ensure that authentication credentials cannot be logged or tampered with during transmission, as they are processed in encrypted form or transformed into non-reversible representations. This intermediary layer maintains access control reliability while preventing credential logging by malicious validators.
Data Source
AI summary
The technology disclosed relates to authenticating users using a plurality of non-deterministic registration biometric inputs. During registration, a plurality of non-deterministic biometric inputs are given as input to a trained machine learning model to generate sets of feature vectors. The non-deterministic biometric inputs can include a plurality of face images and a plurality of voice samples of a user. A characteristic identity vector for the user can be determined by averaging feature vectors. During authentication, a plurality of non-deterministic biometric inputs are given as input to a trained machine learning model to generate a set of authentication feature vectors. The sets of feature vectors are projected onto a surface of a hyper-sphere. The system can authenticate the user when a cosine distance between the authentication feature vector and a characteristic identity vector for the user is less than a pre-determined threshold.