Loading...
Loading...
Contributors:
Dataset URI:
10.5281/zenodo.22681350Measurement artifact for the manuscript "SignSeam: An Empirical Study of Divergent Enforcement of an Optional Protocol Security Feature in PX4 and ArduPilot". MAVLink 2.0 message signing is enforced per channel and per message type, so an implementation may accept selected frames unsigned while signing is active, and two conforming implementations need not leave the same surface. This deposit contains the software-in-the-loop measurement harness, the per-run result records and console transcripts behind every reported number, and a source-level profiler that recovers each stack's unsigned-acceptance surface directly from its source. Contents: per-run JSON records and console transcripts for the ArduPilot diversion campaign (10 flights), for the PX4 campaign and its matched control (10 + 10 flights), the ArduPilot L1 acceptance matrix, the PX4 ingestion runs, the drivers and campaign scripts, and the seam profiler with fixtures and a self-test that runs without cloning either autopilot. Signing is verified active on the ground in every run, and that check is recorded per run rather than assumed. Environment: PX4 at commit bf6f90b with the jMAVSim backend, ArduPilot (ArduCopter) at commit be40f59, headless SITL under WSL2, attacker pymavlink 2.4.49 holding no signing key, each flight booting from a wiped EEPROM. Consistent with the coordinated-disclosure statement in the paper, this is not a generalized or weaponized attack tool; the drivers require a fully configured SITL testbed. The README explains two things needed to read the raw records correctly. First, the takeoff_blocked field is false in every run because no takeoff was commanded after the maneuver: it is a default, not a measurement, and the announced takeoff inhibition is evidenced from the autopilot console transcripts, present in all ten attack runs and in none of the ten controls. Second, the free-text verdict.note annotation, written in Turkish while the runs were being driven, overstates that result; it is left in place rather than edited after the fact, because removing it would hide a discrepancy a reader is entitled to see. Local filesystem paths and account names in logs and scripts were replaced with generic ones; no result record and no measured value was changed.
Keyword(s):
MeSH Term(s):
Grants and Funding:
No Grants or Funding Data Available
Associated Publication(s):
No Associated Publication Data Available
Supplementary Data:
No Supplementary Data Available
Related Dataset(s):
No Related Dataset Data Available
The National Library of Medicine (NLM) provides information about biomedical datasets. Inclusion on an NLM Web page does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: NLM Web Policies (nih.gov)