Generation of a dataset for DoW attack detection in serverless architectures.

Area of Science:

• Cloud Computing Security
• Cybersecurity
• Data Science

Background:

• Denial of Wallet (DoW) attacks exploit serverless environments, causing excessive costs.
• Serverless architectures (Function-as-a-Service) offer scalability but are vulnerable to DoW attacks.
• A lack of realistic datasets hinders research into DoW attack detection.

Purpose of the Study:

• To create a synthetic dataset for simulating function invocations in serverless architectures.
• To facilitate the development and evaluation of machine learning models for DoW attack detection.
• To provide a valuable resource for researchers and developers in serverless security.

Main Methods:

• Generated synthetic data based on real cloud traffic patterns.
• Identified key characteristics of function invocations: SubmitTime, Invocation Delay, Response Delay, Function Duration, Active Functions at Request, and Active Functions at Response.
• Proposed a methodology for dataset generation and its application in attack detection.

Main Results:

• A novel dataset simulating serverless function invocations has been created.
• The dataset captures essential characteristics for analyzing serverless traffic.
• The dataset is available in the Mendeley data repository for public use.

Conclusions:

• The generated dataset is crucial for advancing research in Denial of Wallet attack detection in serverless environments.
• Machine learning techniques and neural networks can be effectively applied using this dataset.
• This resource will enhance the reliability, efficiency, and security of serverless applications.