On-Demand Centralized Resource Allocation for IoT Applications: AI-Enabled Benchmark

Ran Zhang,Lei Liu,Mianxiong Dong,Kaoru Ota

DOI: https://doi.org/10.3390/s24030980

IF: 3.9

2024-02-03

Sensors

Abstract:The development of emerging information technologies, such as the Internet of Things (IoT), edge computing, and blockchain, has triggered a significant increase in IoT application services and data volume. Ensuring satisfactory service quality for diverse IoT application services based on limited network resources has become an urgent issue. Generalized processor sharing (GPS), functioning as a central resource scheduling mechanism guiding differentiated services, stands as a key technology for implementing on-demand resource allocation. The performance prediction of GPS is a crucial step that aims to capture the actual allocated resources using various queue metrics. Some methods (mainly analytical methods) have attempted to establish upper and lower bounds or approximate solutions. Recently, artificial intelligence (AI) methods, such as deep learning, have been designed to assess performance under self-similar traffic. However, the proposed methods in the literature have been developed for specific traffic scenarios with predefined constraints, thus limiting their real-world applicability. Furthermore, the absence of a benchmark in the literature leads to an unfair performance prediction comparison. To address the drawbacks in the literature, an AI-enabled performance benchmark with comprehensive traffic-oriented experiments showcasing the performance of existing methods is presented. Specifically, three types of methods are employed: traditional approximate analytical methods, traditional machine learning-based methods, and deep learning-based methods. Following that, various traffic flows with different settings are collected, and intricate experimental analyses at both the feature and method levels under different traffic conditions are conducted. Finally, insights from the experimental analysis that may be beneficial for the future performance prediction of GPS are derived.

engineering, electrical & electronic,chemistry, analytical,instruments & instrumentation

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to achieve demand - based centralized resource allocation in Internet of Things (IoT) applications, and proposes an artificial intelligence (AI) - supported performance benchmarking method. Specifically, the paper focuses on the impact of different network traffic characteristics on performance prediction under the multi - queue Generalized Processor Sharing (GPS) scheduling mechanism. The paper points out that the existing performance prediction methods have limitations, such as being specific to a certain traffic model (for example, short - range dependence or long - range dependence), which limits their applicability in real - world environments. In addition, the lack of a unified benchmarking method has led to unfair comparisons of performance prediction results. Therefore, this paper designs an AI - supported Traffic - oriented Generalized Processor Sharing (ToGPS) benchmarking, aiming at:

1. **Providing a benchmarking platform**: for evaluating and comparing different performance prediction methods, including traditional approximate analysis methods, traditional machine learning methods, and deep learning methods.
2. **Considering traffic heterogeneity**: combining different traffic models (such as Poisson traffic model and self - similar traffic model) to generate five types of traffic data sets to cover traffic characteristics in different scenarios.
3. **Fairly comparing different methods**: through unified data set formats and evaluation metrics, achieving fair comparison of different performance prediction methods.
4. **Optimizing resource allocation**: through experimental analysis under different traffic conditions, providing insights on how to optimize GPS performance prediction, thereby guiding future resource allocation strategies.

The main contributions of the paper include:
- Providing the first benchmarking platform for performance prediction of the multi - queue GPS scheduling mechanism, making it possible to compare different performance prediction methods in a consistent experimental environment.
- Designing five data sets containing different traffic characteristics, as well as unified data set formats and evaluation metrics, to achieve fair performance prediction comparison.
- Combining knowledge - driven information and machine learning techniques, significantly improving the accuracy of GPS performance prediction.

Through these efforts, the paper provides new perspectives and tools for resource allocation in IoT applications, which helps to improve service quality and network performance.