MSCET: A Multi-Scenario Offloading Schedule for Biomedical Data Processing and Analysis in Cloud-Edge-Terminal Collaborative Vehicular Networks

Zhichen Ni, Honglong Chen, Zhe Li, Xiaomeng Wang, Na Yan, Weifeng Liu, Feng Xia

In the era of autonomous and connected vehicles, your car is becoming more than just a mode of transport—it’s becoming a mobile health clinic. Processing real-time biomedical data (like a driver’s heart rate or fatigue levels) requires massive computational power and ultra-low latency. However, vehicles have limited "brains," the cloud is often too far away, and local roadside units can easily become overwhelmed.

A new paper recently published in IEEE (and available on arXiv) introduces MSCET, a multi-scenario offloading schedule designed to solve this "where-do-I-process-this?" dilemma by orchestrating a three-way collaboration between the vehicle, the edge, and the cloud.

TL;DR

MSCET is a collaborative framework that optimizes how vehicles offload intensive biomedical data tasks across Cloud, Edge, and Terminal (vehicle) layers. By introducing a "virtual resource pool" that aggregates power from multiple roadside units, it ensures driver monitoring remains fast and reliable even in dense, high-traffic environments.

What Problem Does This Paper Address?

Biometric monitoring generates "computation-intensive" and "delay-sensitive" tasks. If the processing takes too long, the safety benefit is lost.

Current vehicular networks face three main hurdles:

  1. Resource Constraints: Vehicles (Terminals) can't handle the heavy lifting of AI-based biomedical analysis.
  2. The Edge Bottleneck: Road Side Units (RSUs) are closer than the cloud, but they have limited capacity. If too many cars offload to one RSU, it crashes or slows down.
  3. Overlapping Coverage: In smart cities, a car might be in range of multiple RSUs. Traditional systems don't know how to efficiently "pool" these overlapping resources.

Methods

The authors propose the MSCET (Multi-Scenario Cloud-Edge-Terminal) schedule. The core innovation lies in how it handles two distinct scenarios:

  • The Overlapping Region: When a vehicle is covered by multiple RSUs, MSCET uses Software Defined Networking (SDN) and Network Function Virtualization (NFV) to create a Virtual Resource Pool. This treats multiple physical edge servers as one large, elastic resource.
  • The General Region: In areas with single RSU coverage, the system uses the Kuhn-Munkres (KM) algorithm to match vehicles to the best RSU based on distance and moving time.

To solve the resulting optimization problem (which is NP-hard), the researchers decoupled the variables and used a hybrid approach:

  • Genetic Algorithms to find the best ratio of data to send to the Cloud.
  • Interior Point Methods to determine how much hardware resource to allocate to each task.
  • KKT (Karush-Kuhn-Tucker) Conditions to analytically solve for the optimal edge-offloading ratio.

Datasets

The paper relies on extensive simulations rather than a static dataset. The simulation environment modeled:

  • A 250-meter road with 5 RSUs.
  • Tasks ranging from 10 to 15 MB with processing requirements of 0.5 GHz per RSU.
  • Vehicle speeds of 40 km/h and local computation resources of 12.5 MHz.
  • Task tolerance delays between 4 and 6 seconds.

Key Insights and Ideas

  • Collaboration, Not Competition: Instead of choosing either Edge or Cloud, MSCET splits a single task into three parts: one processed locally, one at the Edge, and one in the Cloud—all running in parallel.
  • The "Moving Time" Buffer: In general regions, the system accounts for the time it takes a car to reach an RSU. It uses "Cloud-Terminal" cooperation to start processing data while the car is still in transit to the edge's coverage area.
  • Utility over Speed: The model maximizes "System Utility," a weighted metric that balances the provider's profit with the user's Quality of Service (QoS). This ensures the system is economically viable while meeting safety-critical deadlines.

Key Results

  • Rapid Convergence: The optimization algorithm is highly efficient, reaching a stable, near-optimal solution within just 6 iterations.
  • Scalability: Unlike "Nearby" offloading (where cars just talk to the closest RSU), MSCET’s utility remains stable even as the RSU coverage radius increases. This prevents the "overload" effect seen in simpler models.
  • Density Performance: As the number of vehicles increases, MSCET significantly outperforms "Edge-Terminal" only models. When edge resources become scarce, the system intelligently shifts the overflow to the cloud without violating latency constraints.
  • Virtual Pooling Benefits: The virtual resource pool was shown to be critical when local RSU resources are limited, effectively "borrowing" power from neighboring units to maintain performance.

Limitations and Future Work

The authors acknowledge that their current model operates under somewhat idealized conditions. For instance, they assumed constant data transmission rates and stable wireless channels.

Future Work will focus on:

  • Testing the schedule in more volatile, real-world communication environments.
  • Implementing the schedule in a physical testbed to account for hardware-level overhead that simulations might miss.
  • Expanding the "Multi-Scenario" logic to handle more complex road topologies beyond unidirectional paths.

Paper Citation: Zhichen Ni et al., "MSCET: A Multi-Scenario Offloading Schedule for Biomedical Data Processing and Analysis in Cloud-Edge-Terminal Collaborative Vehicular Networks," arXiv:2203.07999v1 [cs.NI], 2022.

Summary updated

Abstract

With the rapid development of Artificial Intelligence (AI) and Internet of Things (IoTs), an increasing number of computation intensive or delay sensitive biomedical data processing and analysis tasks are produced in vehicles, bringing more and more challenges to the biometric monitoring of drivers. Edge computing is a new paradigm to solve these challenges by offloading tasks from the resource-limited vehicles to Edge Servers (ESs) in Road Side Units (RSUs). However, most of the traditional offloading schedules for vehicular networks concentrate on the edge, while some tasks may be too complex for ESs to process. To this end, we consider a collaborative vehicular network in which the cloud, edge and terminal can cooperate with each other to accomplish the tasks. The vehicles can offload the computation intensive tasks to the cloud to save the resource of edge. We further construct the virtual resource pool which can integrate the resource of multiple ESs since some regions may be covered by multiple RSUs. In this paper, we propose a Multi-Scenario offloading schedule for biomedical data processing and analysis in Cloud-Edge-Terminal collaborative vehicular networks called MSCET. The parameters of the proposed MSCET are optimized to maximize the system utility. We also conduct extensive simulations to evaluate the proposed MSCET and the results illustrate that MSCET outperforms other existing schedules.
BibTeX
@article{ni2023mscet,
  author = {Zhichen Ni and Honglong Chen and Zhe Li and Xiaomeng Wang and Na Yan and Weifeng Liu and Feng Xia},
  title = {{MSCET: A Multi-Scenario Offloading Schedule for Biomedical Data Processing and Analysis in Cloud-Edge-Terminal Collaborative Vehicular Networks}},
  journal = {IEEE ACM Trans. Comput. Biol. Bioinform.},
  year = {2023},
  url = {https://openreview.net/forum?id=a8eyJSaHAk},
}
Figure