QoS Aware Mixed-Criticality Task Scheduling in Vehicular Edge Cloud System

TL;DR

The paper tackles online mixed-criticality task scheduling in a Vehicular Edge Cloud System (VECS) to maximize infrastructure profit while minimizing task drop, distance, and energy costs. It proposes a hybrid local/global scheduling framework where AVs offload tasks to nearby BSs when feasible, or to a centralized scheduler when congestion rises, and introduces three scheduling heuristics—Selfish Holding, Nearest Server, and Dynamic Holding—to manage uncertainty and hard-soft task differentiation. The approach is validated on synthetic and real-life datasets, showing up to 25% QoS improvement over baselines and revealing favorable settings for distance and energy trade-offs. The work advances practical VECS provisioning by integrating online mixed-criticality processing, energy-distance awareness, and soft-task dropping into a unified profit-minimization objective, with clear paths for extension to heterogeneous BSs and additional priority levels.

Abstract

Modern-day cars are equipped with numerous cameras and sensors, typically integrated with advanced decision-control systems that enable the vehicle to perceive its surroundings and navigate autonomously. Efficient processing of data from sensors, lidars, radars and cameras is quite computationally intensive and can not be done with good accuracy using less capable onboard resources. In order to deal with this problem, some computation requirements (also referred as tasks) are offloaded to infrastructure or executed in parallel in both autonomous vehicle (AV) and infrastructure to enhance accuracy. The infrastructure comprises base stations, a centralized cloud, and a CS. Base stations (BSs) execute tasks in collaboration with a significantly more powerful centralized cloud, while the centralised scheduler (CS) centrally schedules all the tasks. The base station receives tasks from multiple AVs, each with varying deadlines, criticality, and locations. Our main goal is to maximize the profit of the infrastructure by (a) minimizing the number of drop tasks, (b) minimizing the distance cost for task offloading, and (c) minimizing the energy usage of BSs. In this work, we proposed efficient approaches to schedule the collection of tasks to the BSs, by employing a hybrid scheduling approach where tasks from AVs get allocated to nearby base stations if the nearby BSs are lightly loaded, otherwise AVs send the task to CS for allocation. The CS maximizes the profit by following strategies: (a) selection of BS considering distance and energy consumption, (b) when task load is moderate or low, highly critical tasks run at favourable utilisation, and (c) low-critical tasks are dropped to free up resources for executing high-critical tasks. Based on our experiments, proposed approaches improved the QoS provided by up to 25% compared to the state-of-the-art approach in real-life datasets.

Figures (14)

• Figure 1: System Model
• Figure 2: Order of task arrival, CS scheduling, task migration and execution of tasks in batches
• Figure 3: Tasks $\tau_{i_1}$ and $\tau_{i_2}$ is scheduled in $BS_j$ from time S and F. But at time $D$ task $\tau_{i_2}$ is dropped. The figure shows how the value of variables $x_{i,j}(t),\:y_{i,j}$ and $z_{i,j}$ changes.
• Figure 4: The order of $TOR$, $TOA$, $TSR$, and $TEA$ in VECS is shown. The numbers indicate the time of each event. Figures (a) and (b) indicate local and global modes of operation, respectively.
• Figure 5: The figure depicts the flowchart of our approach. H denotes a component of the VECN, H indicates a decision within the approach, H signifies an event of signal transmission among the VECN components, and H represents the two modes of scheduling discussed in the proposed approach. The sequence of events is marked by numbers. Events in local mode start with "L" (enclosed by blue rectangle), while those in global mode (enclosed by red rectangle) start with "G". The AV first initiates local mode execution; if this is unsuccessful, it then initiates global mode execution. If either execution mode is successful, the process concludes with task offloading to $BS_{select}$.