TSNet-SAC: Leveraging Transformers for Efficient Task Scheduling

Ke Deng,Zhiyuan He,Hao Zhang,Haohan Lin,Desheng Wang

2023-06-16

Abstract:In future 6G Mobile Edge Computing (MEC), autopilot systems require the capability of processing multimodal data with strong interdependencies. However, traditional heuristic algorithms are inadequate for real-time scheduling due to their requirement for multiple iterations to derive the optimal scheme. We propose a novel TSNet-SAC based on Transformer, that utilizes heuristic algorithms solely to guide the training of TSNet. Additionally, a Sliding Augment Component (SAC) is introduced to enhance the robustness and resolve algorithm defects. Furthermore, the Extender component is designed to handle multi-scale training data and provide network scalability, enabling TSNet to adapt to different access scenarios. Simulation demonstrates that TSNet-SAC outperforms existing networks in accuracy and robustness, achieving superior scheduling-making latency compared to heuristic algorithms.

Networking and Internet Architecture,Artificial Intelligence,Machine Learning,Systems and Control

What problem does this paper attempt to address?

The paper mainly targets the issue of task scheduling in autonomous driving systems within the future 6G Mobile Edge Computing (MEC) environment, proposing a new method based on the Transformer architecture—TSNet-SAC. In the 6G MEC architecture, autonomous driving systems need to process a large amount of multimodal data generated by various sensors (such as LiDAR, cameras, radar, etc.), and there is a strong interdependency among these data. However, traditional heuristic algorithms perform poorly in real-time scheduling because they require multiple iterations to find the optimal solution, which does not meet the strict low-latency requirements of autonomous driving. To address the aforementioned problem, the paper introduces TSNet-SAC, which leverages the self-attention mechanism of Transformers to efficiently handle the correlations between data, using heuristic algorithms only to guide the training process. TSNet-SAC consists of two sub-networks: OffloadNet is responsible for task offloading decisions, while ResourceNet focuses on resource allocation. In addition, a Sliding Augmentation Component (SAC) is introduced to enhance the network's robustness and address algorithmic deficiencies, along with an Extender component to handle multi-scale training data, enabling TSNet to adapt to different access scenarios. Through simulation experiments, TSNet-SAC outperforms existing networks and heuristic algorithms in terms of accuracy, robustness, and scheduling latency, especially in scenarios with a fixed number of terminal accesses. This approach provides a new solution for real-time task scheduling in autonomous driving systems within the 6G MEC environment, demonstrating its potential for practical deployment. In summary, the paper aims to solve the real-time and efficiency issues of task scheduling and resource allocation in autonomous driving systems under the 6G MEC environment. By innovatively designing the TSNet-SAC network, it achieves a more optimal scheduling strategy and enhances the overall performance of the system.

TSNet-SAC: Leveraging Transformers for Efficient Task Scheduling

Mayfly Taylor Optimization-Based Graph Attention Network for Task Scheduling in Edge Computing

DeepScheduler: Enabling Flow-Aware Scheduling in Time-Sensitive Networking.

An Online Simulated Annealing-Based Task Offloading Strategy for a Mobile Edge Architecture

Optimizing the Deployment of Tiny Transformers on Low-Power MCUs

Deep Reinforcement Learning With Communication Transformer for Adaptive Live Streaming in Wireless Edge Networks

CA-DTS: A Distributed and Collaborative Task Scheduling Algorithm for Edge Computing Enabled Intelligent Road Network

MAARS: Multiagent Actor–Critic Approach for Resource Allocation and Network Slicing in Multiaccess Edge Computing

SACNet: A Scattered Attention-based Network with Feature Compensator for Visual Localization

Utility-Driven End-to-End Network Slicing for Diverse IoT Users in MEC: A Multi-Agent Deep Reinforcement Learning Approach

Sparse Mobile Crowdsensing for Cost-Effective Traffic State Estimation with Spatio-Temporal Transformer Graph Neural Network

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

Distributed Dependent Task Offloading in CPU-GPU Heterogenous MEC: A Federated Reinforcement Learning Approach

MultiTASC: A Multi-Tenancy-Aware Scheduler for Cascaded DNN Inference at the Consumer Edge

Digital Twin-Driven Collaborative Scheduling for Heterogeneous Task and Edge-End Resource Via Multi-Agent Deep Reinforcement Learning.

Joint Service Deployment and Task Scheduling for Satellite Edge Computing: A Two-Timescale Hierarchical Approach

Optimizing Layer-Fused Scheduling of Transformer Networks on Multi-accelerator Platforms

Efficient Task Migration and Resource Allocation in Cloud–edge Collaboration: A DRL Approach with Learnable Masking

Decomposition-based learning in drone-assisted wireless-powered mobile edge computing networks

M$^3$SAT: A Sparsely Activated Transformer for Efficient Multi-Task Learning from Multiple Modalities

Robust Task Scheduling for Delay-Aware IoT Applications in Civil Aircraft-Augmented SAGIN