Focal Interest
Focal Interest
AI Summary of Peer-Reviewed Research

This page presents an AI-generated summary of a published research paper. The original authors did not write or review this article. [See full disclosure ↓]

Publishing process signals: MODERATE — reflects the venue and review process. — venue and review process.

Hybrid deep learning improved edge-cloud task scheduling in simulation

Multiple blue network cables bundled and connected to server equipment with gold-colored ports and yellow indicator lights in a dark data center environment, viewed from below looking upward.
Research area:Computer ScienceDistributed and Parallel Computing SystemsIoT and Edge/Fog Computing

What the study found

A hybrid actor-critic deep reinforcement learning framework called IntelliScheduler improved task scheduling in an edge-cloud computing setting in simulation. The authors report better performance than the compared baselines on reward, training loss, operational cost, rejection rate, and quality of experience (QoE).

Why the authors say this matters

The authors say edge-cloud computing is important for Internet of Things (IoT) workflow applications because it can support low latency and on-demand resource allocation. They conclude that the adaptive learning formulation is highly relevant for dynamic edge-cloud scheduling scenarios.

What the researchers tested

The researchers developed IntelliScheduler, which combines a runtime-aware state representation, a learning-based decision mechanism, and a multi-buffer experience replay architecture. They also developed a learning-based optimal task scheduling (LbOTS) algorithm to minimize total task execution delay by choosing where tasks should run across edge and cloud resources using latency-aware reward modeling.

What worked and what didn't

In simulation experiments under different workloads, LbOTS reportedly achieved up to 13% higher normalized reward, 67% lower training loss, 52-66% lower operational cost, and 80-90% lower rejection rate than PSO, MBO, and MOPSObaselines. The abstract also states that QoE was approximately 15-75% better. The abstract does not report any results showing the proposed approach performing worse than the baselines.

What to keep in mind

The current assessment was simulation-based, so the abstract does not describe real-world deployment results. The abstract also does not provide detailed limitations beyond that scope.

Key points

  • IntelliScheduler is a hybrid actor-critic deep reinforcement learning framework for edge-cloud task scheduling.
  • The study reports up to 13% higher normalized reward and 67% lower training loss in simulation.
  • Compared with PSO, MBO, and MOPSObaselines, the method reportedly reduced operational cost by 52-66% and rejection rate by 80-90%.
  • The abstract says QoE was approximately 15-75% better.
  • The evaluation was simulation-based, not a real-world deployment test.

Disclosure

Research title:
Hybrid deep learning improved edge-cloud task scheduling in simulation
Authors:
L. Raghavendar Raju, M. Venkata Krishna Reddy, Sridhar Reddy Surukanti, G. Sudhakar, V. V. Subrahmanya Sarma M, Anjaiah Adepu
Institutions:
Mahaveer Academy of Technology and Science University, Chaitanya Bharathi Institute of Technology, Indian Institute of Technology Hyderabad, Koneru Lakshmaiah Education Foundation, Aditya Birla (India), Maulana Azad National Urdu University
Publication date:
2026-02-27
DOI:
10.1038/s41598-026-41330-8
OpenAlex record:
View
AI provenance: This post was generated by OpenAI. The original authors did not write or review this post.

More posts