SOMA-Bench: An Open Synthetic Benchmark and Evaluation Harness for Risk-Aware Recovery and Machine Identities in Post-Quantum IAM

Sravanakumar Nidamanooru

Abstract

Identity and access management (IAM) systems are entering a difficult transition: recovery flows remain attacker-favored, machine identities rotate at scale, and post-quantum cryptography (PQC) introduces larger artifacts and new latency envelopes. Teams need a repeatable way to quantify fraud-versus-friction trade-offs and rollout safety during crypto-agile migrations—without exposing proprietary scoring models. This paper proposes a public, synthetic benchmark and evaluation harness for IAM recovery, sign-in, and credential rotation under PQC-aware conditions. The benchmark contributes (i) event schemas and a configurable generator with knobs for fraud prevalence, distribution drift, and signal dropout; (ii) a PQC “overlay” that models payload sizes and processing overhead for issuance/verification; (iii) simple baseline policies (static MFA, trivial risk); and (iv) reproducible metrics, including fraud blocked (%), legitimate friction (%), p95 decision latency, time-to-innocence, rotation SLO pass rate, and migration health (%C/%H/%Q). We report baseline results and stress tests and release code and documentation to enable independent replication and extensions. This work is the first step in a broader research agenda on SOMA, a risk-aware orchestrator for recovery and machine identities; system internals remain out of scope here and will be detailed in subsequent publications. (A patent application is pending on SOMA’s underlying mechanisms; the benchmark is designed to remain IP-safe while still supporting rigorous comparison.

Keywords

Identity and Access Management (IAM), Post-Quantum Cryptography (PQC), Artificial Intelligence (AI), Risk-Based Authentication (RBA), Crypto agility, Hybrid cryptography, Account recovery, Machine identities, Credential rotation, Risk-based authentication, Evaluation harness, Fraud detection, Legitimate friction, Time-to-innocence (TTI); Migration health (%C/%H/%Q); Drift robustness

Corresponding Author

Sravanakumar Nidamanooru, IAM Specialist, master’s in computer science, Independent Researcher (Identity & Access Management), USA.

Citation

Nidamanooru, S. (2025). SOMA-Bench: An Open Synthetic Benchmark and Evaluation Harness for Risk-Aware Recovery & Machine Identities in Post-Quantum IAM.Cognitive Computing & Extended Realities, 1(1), 01-16.