Deployment and Allocation Strategy for MEC Nodes in Complex Multi-Terminal Scenarios

Danyang Li, Yuxing Mao, Xueshuo Chen, Jian Li, Siyang LiuSensors 2022 · DOI

TL;DR

This paper tackles the challenge of optimally placing and allocating mobile edge computing (MEC) nodes in complex IoT environments with diverse terminal needs. They propose a novel evaluation model considering reward, energy, and cost, and use an improved genetic algorithm (GA) to find the best deployment strategy, outperforming traditional k-means and ant colony algorithms.

What Problem Does This Paper Address?

The Internet of Things (IoT) is exploding, generating massive amounts of data. While cloud computing can handle this, it's often too slow for real-time, latency-sensitive applications. Mobile Edge Computing (MEC) offers a solution by bringing computation closer to the devices. However, deciding where to place these MEC nodes and how to assign tasks to them is crucial for system performance. Existing approaches often fix node locations or overlook critical terminal characteristics like power supply and task urgency. This paper addresses this gap by proposing a comprehensive strategy for MEC node deployment and task allocation in complex, multi-terminal scenarios, considering factors that impact real-world performance.

Methods

The core of this paper is a multi-objective optimization model designed to evaluate MEC node deployment. This model integrates three key factors:

  • Reward: The benefit gained from successfully processing tasks, weighted by terminal urgency.
  • Energy Consumption: The power used by terminals, particularly emphasizing battery-powered devices.
  • Deployment Cost: The expense associated with deploying edge nodes.

To find the optimal solution, the authors employ an improved Genetic Algorithm (GA). They also compare its performance against two other common algorithms:

  • k-means: A clustering algorithm that groups terminals based on distance.
  • Ant Colony Optimization (ACO): An algorithm inspired by ant foraging behavior to find optimal paths.

The GA is chosen for its superior performance in balancing the various objectives. The paper also details how latency, queuing, and transmission times are calculated to feed into the optimization model.

Datasets

Not discussed in the paper. The experiments are based on simulated scenarios with defined parameters for terminals and edge nodes.

Key Insights and Ideas

  • Terminal characteristics matter: Simply placing MEC nodes based on location isn't enough. Factors like a terminal's power source (battery vs. direct power) and the urgency of its tasks significantly influence optimal deployment. High-urgency tasks demand faster processing, which impacts node placement and resource allocation.
  • GA excels in complexity: The Genetic Algorithm proved to be the most effective at balancing the competing objectives of reward, energy consumption, and deployment cost, leading to better overall system performance compared to k-means and ACO.
  • Trade-offs are inherent: The study highlights the inherent trade-offs. For instance, increasing the number of edge nodes can reduce latency but increases deployment cost. The GA effectively navigates these trade-offs.
  • Scenario-specific tuning is key: The performance of the system is highly sensitive to parameters like terminal density, task complexity, and edge node processing power. The paper demonstrates how these factors influence the optimal number of nodes and overall system efficiency.

Key Results

  • The proposed GA-based strategy achieved the highest overall objective function value, indicating the best deployment result when considering reward, energy, and cost.
  • The GA also demonstrated the lowest variance in data distribution, suggesting improved system stability and reduced risk of task blockage.
  • In comparisons, the GA converged faster than ACO and achieved better results than k-means. For example, the GA achieved the best function value after around 300 generations, while ACO took around 600.
  • Experiments showed that increasing terminal density generally requires more edge nodes, but enhancing edge node processing power can reduce the number of nodes needed, leading to cost savings and lower latency.
  • Higher task complexity generally leads to increased latency and a reduction in reward if processing capacity isn't increased, highlighting the importance of matching node capabilities to task demands.

Limitations and Future Work

  • Current scenario limitations: The current model is based on fixed locations for devices and potential edge nodes, which might not reflect dynamic real-world environments.
  • Service type diversity: The paper acknowledges that real-world MEC scenarios involve a wider variety of service types, and future work could integrate planning for these diverse services.
  • Machine learning integration: The authors suggest exploring machine learning approaches for more efficient and simplified strategic allocation of edge nodes.
  • Dynamic deployment: Future research should consider dynamic edge node deployment strategies to adapt to changing network conditions and demands.
  • More comprehensive cost model: The paper aims to develop a more scientific and complete system cost model that better accounts for the nonlinear relationships between task data volume and complexity.

Summary updated

Abstract

Mobile edge computing (MEC) has become an effective solution for insufficient computing and communication problems for the Internet of Things (IoT) applications due to its rich computing resources on the edge side. In multi-terminal scenarios, the deployment scheme of edge nodes has an important impact on system performance and has become an essential issue in end-edge-cloud architecture. In this article, we consider specific factors, such as spatial location, power supply, and urgency requirements of terminals, with respect to building an evaluation model to solve the allocation problem. An evaluation model based on reward, energy consumption, and cost factors is proposed. The genetic algorithm is applied to determine the optimal edge node deployment and allocation strategies. Moreover, we compare the proposed method with the <i>k</i>-means and ant colony algorithms. The results show that the obtained strategies achieve good evaluation results under problem constraints. Furthermore, we conduct comparison tests with different attributes to further test the performance of the proposed method.
BibTeX
@inproceedings{li2022deployment,
  author = {Danyang Li and Yuxing Mao and Xueshuo Chen and Jian Li and Siyang Liu},
  title = {{Deployment and Allocation Strategy for MEC Nodes in Complex Multi-Terminal Scenarios}},
  booktitle = {Sensors},
  year = {2022},
  doi = {10.3390/s22186719},
}
Figure