MoTO-BFT: Intelligent gradient-driven orchestration of BFT-aware controller placement and task offloading in vehicular edge SDN

Software-Defined Networking (SDN) has emerged as a key enabler for intelligent vehicular edge computing by providing centralized control and adaptive resource management. However, achieving reliable controller placement under Byzantine Fault Tolerance (BFT) and efficient task offloading in highly dynamic vehicular environments remains a major challenge due to frequent topology changes, communication delays, and control decision overhead. Existing works treat controller placement and task offloading as independent optimization problems and do not account for reassignment frequency, controller decision time, or BFT-compliant controller mapping. To address these limitations, this paper proposes MoTO-BFT , a novel M ulti- o bjective T ask O ffloading and B yzantine F ault- T olerant controller placement framework for SDN-enabled vehicular networks. MoTO-BFT integrates Temporal Convolutional Network (TCN)-based trajectory prediction with a novel Fast-converging Multi-objective Gradient Aggregation algorithm (FaMOGA) to jointly optimize BFT-compliant controller placement and computation task offloading. The proposed framework minimizes communication and computation delay, controller decision time, controller count, and reassignment frequency while ensuring balanced load distribution and 3 f + 1 BFT controller mapping. Extensive simulations using real vehicular mobility traces demonstrate that MoTO-BFT significantly outperforms existing controller placement and offloading baselines in terms of latency reduction, reassignment stability, and BFT robustness, making it a highly efficient and reliable solution for dynamic vehicular SDN environments.