10/30/2026
By Marley O'Neil

The Francis College of Engineering, Department of Electrical and Computer Engineering, invites you to attend a master's thesis defense by Jonathan Rubio on “Result Handoff in Heterogeneous Byzantine-Resilient System on a Chip (HBR-SoC): A gem5 Evaluation Platform.”

Defense Details

  • Defense date: Friday, Nov. 6, 2026
  • Time: 1–2 p.m.
  • Location: Ball Hall 302

Committee

  • Advisor: Lewis Tseng, Associate Professor, Electrical and Computer Engineering, UMass Lowell
  • Yan Luo, Professor, Electrical and Computer Engineering, UMass Lowell
  • James Moran, Senior Adjunct AY Faculty, Electrical and Computer Engineering, UMass Lowell

Abstract

In a Heterogeneous Byzantine-Resilient System-on-Chip (HBR-SoC), faulty workers can change results during controller voting. Midir and RDMA-based agreement do not compare the latency and buffer-reuse costs of preventing inconsistent reads on a common chip model. This thesis develops a gem5 platform comparing two mechanisms: protected handoff changes memory permissions, while static copy keeps permissions fixed and uses stable copies provided by a trusted monitor. The comparison uses a braking case study in which workers propose acceleration, coasting or braking actions for a cart approaching a fixed boundary. In one deterministic eight-round simulation per configuration, with fixed planner inputs and four 96-byte proposals per round, static copy reduces mean simulated decision time from 236 to 174 microseconds, about 26%, against the fastest tested protected copy configuration. Static copy preserves matching approvals in a mutation test that splits honest votes under unsealed sharing, but it allows delayed controllers to read later buffer contents. With protected handoff, deliberately suppressing an acknowledgment for a controller read grant covering four buffers triggers permanent quarantine, stopping fresh proposals from all four producers despite correct permission enforcement. The platform provides a basis for future studies of spacecraft guidance and robotic navigation, examining how result size, memory contention and recovery policies affect decision latency and continued operation.