REAL-TIME PERFORMANCE OPTIMIZATION OF ELECTRONIC EMBEDDED SYSTEMS USING DEEP REINFORCEMENT LEARNING ALGORITHMS

The rapid evolution of electronic embedded systems (EES) has brought significant challenges in optimizing their performance in real-time environments. These systems are often deployed in critical applications, such as automotive, medical, and IoT devices, where efficient resource management and adaptive decision-making are essential for optimal performance. Traditional optimization methods struggle to meet the dynamic and complex demands of modern embedded systems. As the complexity of electronic embedded systems increases, ensuring real- time performance while minimizing energy consumption, latency, and operational costs becomes more difficult. Static configurations or conventional algorithms cannot adapt quickly to changing conditions, leading to suboptimal performance. This problem is further exacerbated by the need for fast decision-making within limited computational resources. This study proposes using Deep Reinforcement Learning (DRL) algorithms to optimize the real-time performance of electronic embedded systems. DRL leverages an agent- based approach to autonomously learn optimal strategies through trial and error in dynamic environments. The proposed method involves training a DRL model to intelligently manage system resources, adjust parameters, and enhance decision-making in real-time based on feedback from the system’s environment. Key DRL techniques, such as Deep Q-Networks (DQN) and Proximal Policy Optimization (PPO), are utilized to train agents in various system scenarios. The results show that DRL-based optimization significantly improves system efficiency, leading to reduced latency, enhanced throughput, and optimized power consumption without compromising the system’s responsiveness. The proposed method outperforms traditional optimization approaches, particularly in highly dynamic and resource-constrained environments, by enabling continuous adaptation to changing operational conditions.