Venue: IEEE Transactions on Network Science and Engineering, Special Issue
Status: Desk-rejected on scope; appeal sent 2026-04-10. Awaiting outcome.
What it is
The submission argues that AEGIS-style runtime governance can run on resource-constrained edge hardware without breaking the latency or throughput envelopes that edge deployments require. The empirical basis is the AEGIS edge-laboratory benchmark suite — bare-metal performance figures collected on commodity edge hardware against both standard and adversarial workload profiles.
Headline results
From the AEGIS edge laboratory:
| Workload | Throughput | p95 latency | Errors |
|---|---|---|---|
| Edge (standard agent traffic) | 51.8 RPS | 29.3 ms | 0 / 5,000 |
| Adversarial (ATX-1-shaped) | 15.1 RPS | 109.8 ms | 0 / 5,000 |
These figures were collected as part of the Round 1 Edge Deployment Evaluation and are bridged into the paper as the Condition-B performance overhead evidence.
Submission status
- 2026-04-10 — Appeal sent to the editor and seven Special Issue guest editors after the initial desk rejection on scope grounds.
- Pending — outcome of the appeal.
Pre-submission caveat
The original ADR-0011 hypothesis used “sub-millisecond” latency framing. Observed bare-metal numbers are 18–66 ms median. The hypothesis language needs correction in the paper to match observed data before any further submission. This correction is held with the revision.
Relationship to other AEGIS work
- The Cross-Cutting Runtime Enforcement paper shares the architectural framing.
- Round 1 provides the benchmark numbers.
- The forthcoming Round 2 — AoC Replication will provide live-load Condition-B performance data that extends the bare-metal figures.