Trustworthy and efficient project scheduling in IIoT based on smart contracts and edge computing

Peng Liu, Xinglong Wu, Yanjun Peng, Hangguan Shan, Saïd Mahmoudi, Bong Jun Choi, Haksrun LaoJ. Cloud Comput 2025 · DOI

Introduction

In the era of Industry 4.0, "smart factories" rely on modular robotic agents to execute complex manufacturing tasks. However, managing these resources across multiple concurrent projects presents a significant challenge known as the Resource-Constrained Multi-Project Scheduling Problem (RCMPSP). This problem is mathematically NP-hard, requiring sophisticated algorithms to find efficient schedules. Furthermore, in decentralized environments involving multiple stakeholders, ensuring the trustworthy execution of these schedules is paramount. This paper proposes a novel framework combining edge computing, a specialized auction algorithm, and blockchain technology to address these hurdles.

System Architecture and Workflow

The proposed system utilizes a three-layered approach to balance computational efficiency with security. The workflow begins with the customer selecting a factory via a Decentralized Application (Dapp), which generates a smart contract.

Overview of the proposed smart factory scheduling workflow

As illustrated in the workflow above, heavy computational tasks are offloaded to edge servers. These servers calculate feasible project scheduling plans and transmit them back to the smart factory. The AIRA (Autonomous Intelligent Robot Agent) platform serves as the bridge between the physical factory operations and the blockchain, ensuring that customer demands are uploaded and results are validated before payments are released.

The CA-RS Algorithm

At the heart of the scheduling engine is the Combinatorial Auction with Random Sampling (CA-RS) algorithm. This method treats projects as bidders competing for robot time slots.

  • Dynamic Pricing: To resolve resource conflicts, the algorithm increases the "price" of a resource when multiple projects demand it simultaneously, naturally incentivizing a balanced load across the factory.
  • Random Sampling: Unlike rigid priority rules (e.g., "Shortest Duration First"), CA-RS employs random sampling based on priority probabilities. This allows the algorithm to explore the solution space more effectively and avoid getting trapped in suboptimal local minima.

Blockchain Integration and Trust

To bridge the "trust gap," the framework integrates the AIRA platform with the Ethereum blockchain. This ensures that every task assignment is immutable and traceable.

Sequence diagram of the AIRA-based smart contract interaction process

The interaction process follows three distinct phases as shown in the sequence diagram:

  1. Negotiation: Demands and offers are matched by a provider node, leading to the creation of a new "liability" on the blockchain.
  2. Execution: The AIRA platform monitors the robot's progress, triggering events and passing arguments necessary for task completion.
  3. Finalization: Once the result is produced, it is finalized on-chain, and the smart contract automatically handles the transfer of funds. This decentralized approach eliminates the need for human intermediaries and protects against "imposture" attacks.

Performance Evaluation

The researchers evaluated the CA-RS approach using the RCMPSPLIB (j609_1 instance), featuring 4 projects and 60 tasks each.

  • Efficiency: The CA-RS method achieved an optimal completion time ($C_{max}$) of 368 units. Remarkably, it reached this optimum using only approximately 7% of the iterations required by a standard Genetic Algorithm (GA).
  • Feasibility: Simulations confirmed that resource usage remained within strict supply capacities for all four machine types, proving the algorithm's real-world applicability.
  • Security: Analysis confirmed that the blockchain-backed system effectively ensures faithful execution and prevents unauthorized tampering with the scheduling plans.

Conclusion and Future Directions

The integration of CA-RS with edge computing and blockchain provides a robust solution for IIoT resource management. While highly effective for static demands, future work will focus on handling "stochastic" or unpredictable order statuses using AI-driven predictive modeling. Additionally, the researchers aim to explore proactive caching strategies to further reduce latency in edge-to-factory communications.

Summary updated

BibTeX
@article{liu2025trustworthy,
  author = {Peng Liu and Xinglong Wu and Yanjun Peng and Hangguan Shan and Saïd Mahmoudi and Bong Jun Choi and Haksrun Lao},
  title = {{Trustworthy and efficient project scheduling in IIoT based on smart contracts and edge computing}},
  journal = {J. Cloud Comput.},
  year = {2025},
  doi = {10.1186/s13677-025-00726-z},
}
Figure