VANET Healthcare Communication Dataset for DRL-Based Priority-Aware Scheduling

This dataset comprises 30,000 simulated network event records generated to model heterogeneous healthcare data transmission in a Vehicular Ad Hoc Network (VANET) environment supporting emergency medical services. Each record captures a single packet transmission event across a dynamic vehicular network involving three node types — ambulances, standard vehicles, and roadside units (RSUs) — reflecting realistic V2V (vehicle-to-vehicle) and V2I (vehicle-to-infrastructure) communication topologies. The dataset contains 18 attributes spanning vehicle mobility parameters, wireless channel conditions, healthcare data characteristics, scheduling decisions, and network performance outcomes. Vehicle positions are represented as 2D Cartesian coordinates within a simulated urban grid, with speeds ranging across realistic mobility profiles. Channel quality is captured through link stability scores and available bandwidth spanning 5–30 Mbps, representing the highly dynamic and intermittent connectivity typical of vehicular environments. Healthcare data is categorized into three clinically motivated types — ECG signals, vital signs, and emergency alerts — each assigned a priority level of 1 (critical), 2 (urgent), or 3 (routine), directly modeling the heterogeneous urgency structure described in the proposed framework. Transmission deadlines and data sizes reflect the real-time delivery constraints of each data class. Network performance is recorded through end-to-end packet delay (10–300 ms), packet loss rate, and allocated bandwidth per transmission event. The scheduling decision column encodes one of five actions — immediate transmit, allocate high bandwidth, compress, queue, and drop low priority — representing the heuristic scheduler's output under varying network and priority conditions. A reward signal aligned with the Deep Q-Network (DQN) training objective is included, computed as a function of latency, reliability, and deadline compliance, enabling direct use of the dataset as an offline experience replay source for reinforcement learning. This dataset directly supports implementation and evaluation of the proposed framework's three core components: the priority-aware heuristic scheduler (Section 3.3), the PSO-based bandwidth allocator (Section 3.4), and the DQN-based routing and scheduling policy (Sections 3.5–3.7). Dataset Specifications Field Detail Total Records 30,000 Total Features 18 File Format Microsoft Excel (.xlsx) File Size ~5.77 MB Node Types Ambulance, Car, RSU Healthcare Data Types ECG, Vital Signs, Emergency Alert Priority Levels 1 = Critical, 2 = Urgent, 3 = Routine Scheduling Actions 5 (immediate transmit, allocate high bandwidth, compress, queue, drop low priority) Bandwidth Range 5.0 – 30.0 Mbps Delay Range 10.0 – 300.0 ms Avg Packet Loss Rate ~9.98% Avg Reward Signal 1.067 Simulation Type Synthetic VANET urban mobility simulation