RL-Based CEP Operator Placement Method on Edge Networks Using Response Time Feedback.

The placement of operators in Complex Event Processing (CEP) services, handling real-time data with DAGs, faces challenges due to the NP-hard nature and edge environment complexity. Prior research by Cai et al. used a predictive and greedy approach to minimize delay during placement, but it degrades with increased node count or event rate fluctuations. We propose a novel approach for CEP operator placement using response time feedback to adapt to the dynamic edge environment. We formulate the problem as a Markov Decision Process and use reinforcement learning for optimal policy learning. Our objective is to minimize total response time in edge environments. Extensive simulations evaluate our approach, which outperforms the greedy method with a 25% average reduction in response time.