Edge computing task scheduling method based on user's social relations: a construction and solution for Smart City Library

Yun Teng, Zijia LiuPeerJ Comput. Sci 2024 · DOI

TL;DR

This paper introduces a novel edge computing task scheduling method for Smart City Libraries that leverages user social relationships to optimize task offloading. By constructing a social relationship graph and a device connection graph, the method aims to minimize task execution delay and energy consumption, leading to improved user experience.

What Problem Does This Paper Address?

The rapid growth of mobile devices and real-time applications in Smart City Libraries has strained existing computing resources, leading to subpar user experiences. Edge computing offers a solution by delegating tasks to edge devices, but challenges remain in user reluctance, device resource constraints, and ensuring secure and reliable task offloading. Specifically, this paper addresses the need for a task scheduling mechanism that considers not only technical parameters but also the social dynamics between users and devices to achieve efficient and trustworthy task execution.

Methods

The authors propose a multi-stage approach:

  1. Social Relationship Graph Construction: They quantify user social relationships using mathematical convolution and the Jaccard similarity ratio. This involves calculating service assessment factors, recommendation trust factors, and similar recommendation factors to build a comprehensive social relationship graph.
  2. Device-Service Connection Graph Construction: This graph is built using Euclidean distance to determine the physical proximity between user equipment and service devices, ensuring compliance with Device-to-Device (D2D) communication conditions.
  3. Task-Device Bipartite Graph Construction: The social relationship graph and the device connection graph are combined to create a task-device bipartite graph. Edge weights in this graph are determined by task execution costs.
  4. Task Scheduling Model: A model is formulated that considers task execution delay, energy consumption of task execution, and user equipment standby energy consumption.
  5. Kuhn-Munkres (KM) Algorithm: The KM algorithm is employed to solve the task scheduling problem, aiming to minimize the overall execution delay and energy consumption.

Datasets

The paper mentions using "Dataset 5 collected from a taxi-hailing service program in 2022," which includes passenger and driver information. This dataset was used for experimental analysis.

Key Insights and Ideas

  • Social Capital in Edge Computing: The core idea is that social relationships between users and devices can be a valuable factor in optimizing task scheduling, moving beyond purely technical metrics. This can lead to more trusted and reliable offloading decisions.
  • Quantifying Trust: The paper demonstrates a method to quantify "social trust" through a combination of direct assessments, recommendations from intermediaries, and similarity with other users' assessments. This provides a structured way to incorporate trust into scheduling.
  • D2D for Efficiency: Leveraging D2D communication is highlighted as a key enabler for efficient task offloading, offering higher link gain and reduced energy consumption compared to traditional network communication.
  • Holistic Optimization: The proposed method considers multiple objectives simultaneously: minimizing delay, minimizing energy consumption, and ensuring task safety and reliability, which is a more comprehensive approach than many existing methods.

Key Results

The proposed KMWSR (KM with social relations) algorithm was compared against three other algorithms: Pure KM (PKM), improved Genetic Algorithm (IGA), and JPORA.

  • Objective Function: KMWSR consistently achieved the best objective function values, indicating a better balance of delay and energy consumption.
  • Task Execution Delay: While JPORA showed slightly better performance in reducing task execution delay in some scenarios, KMWSR demonstrated more stable performance across varying numbers of devices and tasks, especially for larger task volumes.
  • Energy Consumption: KMWSR was the most effective in terms of energy saving, outperforming other algorithms, particularly as the number of mobile devices increased.
  • Social Threshold Impact: KMWSR was least affected by changes in the social relationship threshold, showing more stable performance and preventing task rejection by leveraging social connections.

Limitations and Future Work

  • Social Threshold Dynamic Adjustment: The paper suggests that in practical scenarios, the edge server should dynamically adjust the social relationship threshold based on real-time user social relationships for optimal performance.
  • Generalizability: While tested on a taxi-hailing dataset, the applicability to other Smart City Library scenarios or different types of tasks could be explored further.
  • Complexity of Social Factors: The paper focuses on specific quantifiable social factors. Future work could explore a wider range of social dynamics and their impact on task scheduling.

Summary updated

BibTeX
@article{teng2024edge,
  author = {Yun Teng and Zijia Liu},
  title = {{Edge computing task scheduling method based on user's social relations: a construction and solution for Smart City Library}},
  journal = {PeerJ Comput. Sci.},
  year = {2024},
  doi = {10.7717/peerj-cs.2457},
}
Figure