MAPS‐PBFT: Mobility‐Aware Persistent‐Survivor PBFT for Vehicular Networks

Abstract

PBFT provides deterministic finality under Byzantine faults but assumes stable committee membership, an assumption that breaks in vehicular networks where mobility‐driven churn can render committee members unreachable mid‐protocol. Existing IoV adaptations filter participants by trust or reputation, or modify the protocol's internal structure, yet none assess whether a selected committee will survive the full consensus execution window. We propose MAPS‐PBFT, a mobility‐aware orchestration layer around unmodified PBFT. The framework explicitly partitions the fault budget into a Byzantine slice be$$ {b}_e $$ and a mobility‐loss slice he$$ {h}_e $$ , treating mobility‐induced unreachability as crash‐equivalent failure. Committee selection is formulated as a survivability‐constrained problem over a logical multi‐hop committee graph: candidates are scored by beacon freshness, multi‐hop communication delay, predicted displacement, and inter‐vehicle proximity, then assembled via greedy incremental construction that rewards intra‐committee reachability. A model‐based launch score, derived from a joint formulation of mobility and Byzantine‐membership risk, gates each consensus instance through a user‐configurable threshold. To make the scope of our claims explicit, we distinguish components that are theoretically developed (the joint Monte Carlo launch score, the mobility‐loss budget, and the view‐change–inclusive execution horizon), fully implemented (mobility‐aware committee selection and epoch orchestration around unmodified PBFT in a full IEEE 802.11p co‐simulation), and approximated in the present experiments (a deterministic proxy for the candidate score, with the launch gate evaluated post hoc and view changes disabled). A 756‐run co‐simulation using OMNeT++, INET, Veins, and SUMO across three topologies and four vehicle densities shows that MAPS outperforms six baselines, improving average success rate by +30.7% over random, +8.8% over nearest, and +10.2% over connectivity‐only selection, with advantages that widen as committee size grows. Post hoc threshold analysis confirms that the launch score discriminates between viable and non‐viable committees, raising conditional success from 76.0% to 92.8% at a moderate gate threshold.