Performance analysis of data transmission in industrial IoT for multi-state sensor nodes

Abstract

In industrial IoT data acquisition systems, sensor nodes are typically deployed in harsh environments and operate under stringent energy and reliability constraints. Unlike the idealistic assumption of continuous activity in most existing transmission studies, real-world nodes exhibit multi-state behavior due to battery depletion, duty-cycle scheduling, and hardware degradation, encompassing four states: active, sleep, faulty, and relay. This paper addresses this practical issue by proposing a transmission performance analysis method that integrates a multi-state model of sensor nodes. First, a four state node model is established, and the functional relationship between the effective arrival rate and the node state probabilities is derived. Second, a state transition matrix is introduced to characterize the dynamic evolution of node states, and the node state probabilities are incorporated into an M/M/1/K queueing system, leading to closed-form expressions for system throughput, packet loss rate, and delay. Finally, two simulation experiments are conducted to validate the effectiveness of the proposed model. The first experiment verifies that under static steady-state conditions, system performance depends only on the active probability and is independent of the granularity of state subdivision. The second experiment reveals the long-term performance degradation induced by node state transitions: within 200 hours, system throughput decreases by 68.6%, while delay increases by 60.9%. In contrast, the conventional two-state model, which assumes constant state proportions, fails entirely to capture this dynamic degradation trend. The findings of this paper provide theoretical foundations for node deployment, duty-cycle configuration, and long-term operational optimization in industrial IoT systems.