MAFAF: Mobility-Aware Flow Anchoring and Dynamic State Migration in Edge SDN Using P4

Abstract

The growing mobility of IoT devices and the stringent latency requirements of edge applications present significant challenges to achieving seamless handovers, scalability, and security in Software-Defined Networking (SDN) environments. Traditional controller-centric mobility mechanisms introduce excessive handover latency, signaling overhead, and session disruptions. This paper proposes the Mobility-Aware Flow Anchoring Framework (MAFAF), a P4-based architecture that embeds flow anchoring and dynamic state migration directly into the programmable data plane, enabling mobility handling without tunneling or constant controller involvement. The framework was evaluated using Mininet with BMv2 software switches for real-world emulation, while scalability assessments were conducted via simulated Intel Tofino hardware. Results show that MAFAF achieves handover latency of 7.8 ± 0.6 ms using digest-triggered migration and 4.3 ± 0.4 ms with proactive migration, maintains session continuity rates of 96.5% for TCP and 98.2% for UDP, and limits packet loss to under 1.1% under moderate load. Simulated Tofino-class hardware supports over 25,000 concurrent flows with only 65% register utilization, confirming the framework’s scalability. Security mechanisms implemented within the P4 pipeline including AnchorID validation, timestamp-based replay protection, and per-device flow quotas achieved spoofing and replay attack detection rates of ≥98.6% with a false positive rate below 1.2%. These results validate the hypothesis that in-switch flow intelligence can reduce handover latency below 10 ms, sustain session continuity above 95%, and ensure secure mobility handling with minimal controller overhead, making MAFAF a robust and efficient solution for next-generation edge SDN–IoT deployments.