P4-TAP: A Data Plane Framework for Traffic Aggregation and Prioritization in Time-Sensitive IoT Networks

Abstract

The explosive growth of the Internet of Things (IoT) has intensified the need for intelligent traffic management capable of reconciling latency-critical and non-critical data streams. This study proposes P4-TAP, a P4-Enabled Traffic Aggregation and Prioritization framework that operates entirely within the data plane to deliver real-time responsiveness for critical IoT traffic while improving bandwidth efficiency for routine communications. The research addresses four measurable objectives: (1) minimize end-to-end latency and jitter for time-sensitive packets, (2) maximize packet rate reduction and aggregation ratio for non-critical traffic, (3) ensure scalability at the edge, and (4) validate performance reproducibility. Implemented on a Mininet-based P4 Behavioral Model v2 (simple_switch) testbed comprising four hosts and two switches, the framework was evaluated under controlled mixed-traffic scenarios involving up to 97,656 packets/s. Results show that P4-TAP reduces the end-to-end latency of critical traffic by 94.4 ± 0.6% and jitter by 94.7 ± 0.5%, while the Stateful Aggregation Logic (SAL) achieves a packet rate reduction of 92.8 ± 0.3% and an average aggregation ratio of 14:1. Unlike prior P4 or SDN-based solutions limited to static prioritization or centralized control, P4-TAP introduces a dual-path in-data-plane architecture that dynamically separates and optimizes critical and non-critical flows in real time. Experimental profiling indicates a switch register utilization below 8% and timer load under 3%, confirming scalability on commodity hardware. These findings validate P4-TAP as a robust and scalable data-plane framework for time-sensitive IoT environments.