Application of Genetic Algorithm-Grey Wolf Optimization-Support Vector Machine Algorithm in Network Security Services Assessment and Prediction

Guoying Han, Bin Zhou, Yazi ZhangJournal of Cyber Security and Mobility 2024 · DOI

TL;DR

This paper introduces a robust two-stage framework for evaluating and predicting network security posture. By combining spectral clustering for assessment with a hybrid GA-GWO-SVM model for prediction, the authors achieved significantly lower error rates (MAPE of 0.0270) and faster processing times compared to traditional optimization methods like Particle Swarm or Artificial Bee Colony algorithms.

What Problem Does This Paper Address?

As the volume of global network traffic explodes, so does the complexity of cyber-attacks. Traditional Network Security Service (NSS) assessment often relies on rigid, rule-based systems or expert knowledge, which struggle to adapt to the dynamic nature of modern threats.

Existing machine learning approaches, such as Support Vector Machines (SVMs), offer a path forward but face a "Goldilocks" problem: they are highly sensitive to their internal parameters. If parameters are tuned poorly, the model fails to generalize; if tuned using standard optimization algorithms (like PSO), they often get stuck in "local optima"—finding a good solution but missing the best one. The authors aim to bridge this gap by creating a more accurate, globally optimized prediction engine.

Methods

The researchers proposed a dual-phase approach:

  1. Assessment via Spectral Clustering: Instead of simple grouping, the study uses spectral clustering (a graph-partitioning method) to quantify the current security state. This method is particularly effective at uncovering hidden abnormal data patterns within complex network traffic.
  2. Prediction via GA-GWO-SVM: This is the core technical innovation.
  3. SVM: Acts as the primary engine for classification and prediction.
  4. Grey Wolf Optimization (GWO): Simulates the leadership hierarchy and hunting mechanisms of gray wolves to find the optimal kernel parameters for the SVM.
  5. Genetic Algorithm (GA): GWO can sometimes suffer from slow convergence. The authors integrated GA to enhance the "global search" capability, ensuring the algorithm explores the entire solution space before the "wolves" begin their refined hunt.
  6. Circle Chaotic Map: Used during the initialization phase to ensure a diverse starting population, preventing the model from becoming biased early on.

Datasets

The study utilized the CIC-IDS2017 dataset, a benchmark created by the Canadian Institute for Cybersecurity. This dataset is comprehensive, containing a mix of normal traffic and common attack vectors, including:

  • Denial-of-Service (DoS)
  • Distributed Denial-of-Service (DDoS)
  • Malware

The researchers used this data to train the models and then compared their results against several baselines: GA-SVM, PSO-SVM (Particle Swarm), and ABC-SVM (Artificial Bee Colony).

Key Insights and Ideas

  • The Power of Hybridization: The "aha moment" here is that GA and GWO are complementary. GA is excellent at broad exploration (finding the right neighborhood), while GWO excels at exploitation (finding the exact house in that neighborhood). Combining them eliminates the weaknesses of using either in isolation.
  • Efficiency in Assessment: The authors found that spectral clustering wasn't just more accurate than the industry-standard K-means; it was also twice as fast, maintaining an assessment time below 0.3 seconds.
  • Quantifying Security: The paper provides a clear mathematical framework for quantifying "security posture" by weighing factors like asset importance, vulnerability severity, and attack frequency into a single numerical index.

Key Results

The proposed GA-GWO-SVM model outperformed all baseline methods across the board:

  • Lower Error: The Mean Absolute Percentage Error (MAPE) for the proposed model was 0.0270, compared to 0.0745 for PSO-SVM and 0.0952 for ABC-SVM.
  • Stability: The Root Mean Square Error (RMSE) was the lowest among tested models at 0.0130, indicating high prediction stability.
  • Maximum Error Reduction: The maximum error was capped at 0.4112, a significant improvement over the standard GA-SVM's 0.5896.
  • Speed: The assessment phase was completed in roughly 0.3s, whereas K-means took approximately 0.6s.

Limitations and Future Work

The authors are transparent about the "speed vs. accuracy" trade-off. While the GA-GWO-SVM model is highly precise, the complexity of the triple-algorithm fusion makes it "slightly insufficient" in terms of raw computing speed for ultra-high-speed, real-time environments.

Future research will focus on optimizing the algorithm's efficiency to reduce computational overhead, potentially making it viable for edge computing or real-time deep packet inspection at the ISP level.

Summary updated

Abstract

The continuous development of information technology has also promoted the progress of the Internet. More people are joining the Internet. The amount of data stored in the network is also increasing, including some important information, which leads to criminals launching attacks on network security. In order to solve the large error in network security situation assessment and poor progress in network security prediction, the study uses spectrum clustering analysis to evaluate the network security situation. Then genetic algorithm, grey wolf optimization algorithm and support vector machine fusion algorithm are used to predict the Network Security Service (NSS). The genetic algorithm is used to optimize the global search ability of the gray wolf optimization algorithm and the parameters of the support vector machine are optimized to evaluate and predict the NSS. The results showed that the maximum error of the proposed model was 0.4112, and the maximum error was 0.5896. The absolute percentage error of this algorithm was 0.0270, while the other algorithms were 0.0745 and 0.0952, respectively. The proposed model has lower errors and time consumption in training and simulation testing compared with other current methods. The network situation assessment and prediction method proposed in the study can effectively improve network security services, ensure the personal information security, and enhance the security of the Internet.
BibTeX
@article{han2024application,
  author = {Guoying Han and Bin Zhou and Yazi Zhang},
  title = {{Application of Genetic Algorithm-Grey Wolf Optimization-Support Vector Machine Algorithm in Network Security Services Assessment and Prediction}},
  journal = {Journal of Cyber Security and Mobility},
  year = {2024},
  doi = {10.13052/jcsm2245-1439.1356},
}
Figure