Enhancing NOMA User Rates in Smart Train Communication Systems via Joint RIS-IOS Deployment: Cascaded Channel Estimation and Phase Shift Design Under Impulsive Noise

Abstract

In this paper, a novel framework is proposed to enhance the performance of power-domain non-orthogonal multiple access (PD-NOMA)-based communication systems in smart trains. The framework leverages the simultaneous deployment of reconfigurable intelligent surface (RIS) and intelligent omni surfaces (IOS) to improve users’ data rates. Due to the presence of impulsive noise arising from both internal and external electromagnetic sources within the train environment—and considering the high sensitivity of PD-NOMA systems to such disturbances—accurate estimation of the cascaded channel comprising both direct and reflected paths among the transmitter, RIS/IOS, and users is of critical importance. To address this, a correntropy-based stochastic gradient ascent (CSGA) algorithm is developed to provide robust channel estimation. Subsequently, a joint phase design strategy for the RIS and IOS is proposed, aiming to maximize the sum rate of PD-NOMA users. Simulation results demonstrate that the CSGA algorithm yields significantly higher channel estimation accuracy compared to the conventional stochastic gradient descent (SGD) approach. At SNR = 15 dB, CSGA algorithm reduces the mean square deviation (MSD) by approximately 12.2 dB and improves the sum-rate by about 2.59 b/s/Hz. The proposed framework, by employing joint phase optimization of RIS and IOS, exhibits strong resilience to impulsive noise and outperforms conventional architectures.