Automated Discovery of CEP Applications with Evolutionary Optimization

Abstract # ↰

Complex Event Processing (CEP) emerged as a powerful tool for detecting key patterns in modern digital systems like smart grids or vehicular networks. CEP platforms like Apache Flink CEP have facilitated its adoption, decoupling application semantics from distributed/parallel execution across cloud-to-edge infrastructures. However, a barrier to wider adoption remains. While system experts can identify meaningful event patterns, they often lack the programming skills required to implement CEP applications, a gap hindering the effective use of CEP systems. This work presents preliminary studies on leveraging evolutionary computation to bridge this gap with automatic CEP application discovery from data. Our approach allows system experts to provide examples of relevant event sequences, from which an evolutionary algorithm can generate CEP applications capable of detecting similar patterns. We show initial results are promising, and several CEP-related challenges can stimulate research in this context.