Analysis of Delay in Internet of Things Enabled Network: A State Space Control Perspective

Abstract

This paper presents a novel control framework designed to address critical challenges of delay and packet loss in networked control systems using edge-cloud integration. The proposed solution utilizes a hybrid approach combining a predictive state-space proportional (PSSP) controller and an adaptive state-space proportional-integral (ASSPI) controller. The predictive controller uses a Luenberger observer to estimate system states, enabling it to proactively compensate for network-induced delays and ensure real-time stability. The adaptive PI controller, a core component of the framework, employs a machine learning algorithm based on Soft Actor-Critic (SAC) with a Long Short-Term Memory (LSTM) network to predict future system states and dynamically adjust control parameters. This intelligent adaptation is further enhanced by integrating Recursive Least Squares (RLS) and an Unscented Kalman Filter (UKF) with an adaptive pole placement technique, providing robust, real-time parameter identification and accurate state estimation. The combined approach demonstrates superior performance in mitigating the effects of variable network latency and data loss, offering a reliable control strategy for latency-sensitive applications. The results show that this integrated compensator is highly effective, paving the way for more resilient and high-performance edge-cloud control systems.