BIMS Remediation Study

Biometix Pty Ltd

1. Architectural Considerations

• Re-platforming Options:
  Consider migrating to an interoperable and scalable framework (e.g., MOSIP) to ensure a sustainable long-term solution. Re-evalute business processes is recommended.
• Operational Efficiency:
  Optimize resource utilization, reduce the number of licenses, and adopt a scalable architecture to lower operating and maintenance costs.
• Alternative Approaches:
  Explore open-source solutions or integrate a Commercial-Off-The-Shelf (COTS) ABIS.

New Field Server

(largely compatible with

BIMS central)

keeps existing frontend

2. Biometric De-duplication Optimisation

• Current Process Inefficiencies
  • ​The simultaneous matching of iris and fingerprint biometrics leads to unnecessary computational overhead and longer processing times 
• Sequential Matching Strategy
  • ​Implement a sequential approach—such as prioritizing iris matching—to streamline the workflow and reduce resource consumption without compromising accuracy.

3. System Testing and Performance Optimisation

• Automated Testing Framework
  • ​The absence of a robust automated testing framework limits the ability to validate software updates and detect performance regressions.
• Synthetic Data & Load Testing
  • ​Utilize synthetic data generation and load testing to simulate real-world conditions and verify system resilience under high operational loads.
• Enhanced Logging and Audits
  • ​Improve logging capabilities and conduct regular system performance audits for continuous monitoring.

4. Opportunities for Face Matching Integration

• Additional Biometric Modality
  • ​ Deployment may require improvements in enrollment quality controls
• Hybrid Verification Model:
  • ​Develop a model where face matching serves as a complementary verification method alongside fingerprint and iris matching.
  • Might allow new use cases
• Quality and Testing
  • Requires more emprical field data

5. Data Merge Strategy

• Current Merging Challenges:
  The existing processes for merging biometric data across multiple encounters can lead to inconsistencies and reduced match efficacy.
• Proposed Improvements:
  Enhance data merging by balancing recency, completeness, and quality to ensure consistent and accurate matching.

 Conclusions and Next Steps

Note: See Appendix 6.4 for proposed assistance in implementing the next steps.

• Conclusions
  • ​Existing matching solution is overly complex, with too many free parameters and ways it can fail
  • It needs a radical overhaul to:
    • simplify management of quality and accuracy
    • improve logging and auditing
    • support testing at scale
    • create an easy deployment pathway
    • remove as much mainentance as possible from UNCHR
    • support synchronisation (out of the box)
  • whilst
    • maintaining all critical features
    • being much more cost effective

Conclusions and Next Steps

• Long-Term Improvements:
  • Refine the cascaded matching workflow (with iris matching prioritized).
  • Evaluate and implement re-platforming options to achieve a more cost-effective and scalable system.

Note: See Appendix 6.4 for proposed assistance in implementing the next steps.

• Immediate Priorities:
  • Look at replacement matching algorithms to reduce field server cost
  • Consider a new testing framework to support both load and regression testing.
  • Conduct empirical studies on sequential matching and also face recognition.