Enhancing Cybersecurity Through Fusion of Optimization With Deep Wavelet Neural Networks on Denial of Wallet Attack Detection in Serverless Computing

A Denial of Wallet (DoW) attack is a cyberattack designed to drain financial resources by causing excessive charges or costs on a serverless computing or cloud platform. This attack is mainly associated with serverless frameworks due to features like auto-scaling, cost amplification, pay-as-you-go, and limited control. Serverless computing, or Function-as-a-Service (FaaS), is a cloud computing (CC) model that enables developers to build and run applications without managing traditional server infrastructure. Deep learning (DL) models show strong potential in detecting DoW attacks, where attackers disrupt services by exploiting system resources. By utilizing advanced neural network architectures, DL methods can identify complex patterns in network traffic, offering enhanced accuracy and resilience against such attacks. This study presents a Fusion of Optimization with Deep Wavelet Neural Networks on Denial of Wallet Attack Detection (FODWNN-DoWAD) approach in serverless computing. The main intention of the FODWNN-DoWAD approach is to enhance cybersecurity in DoW attack detection. The FODWNN-DoWAD method utilizes min-max normalization to scale the input network data into a beneficial format. Besides, the pair barracuda swarm optimization (PBSO) method is employed to select features optimally. To recognize a DoW attack, the FODWNN-DoWAD method utilizes a deep wavelet neural network (DWNN) model. Finally, the hyperparameter tuning of the DWNN classifier is performed using the hierarchical learning-based chaotic crayfish optimizer (HLCCO) model. Wide-ranging experiments are accomplished on a benchmark database to demonstrate the good classification outcome of the FODWNN-DoWAD model. The performance validation of the FODWNN-DoWAD approach portrayed a superior accuracy value of 99.05% over existing methods.