Analysis and Optimization of Multi-Active RIS Assisted Cell-Free Massive MIMO

This paper investigates the sum spectral efficiency (SE) and total energy efficiency (EE) of the active reconfigurable intelligent surface (aRIS)-assisted cell-free (CF) massive multiple-input multiple-output (mMIMO) over temporal and spatially correlated channels. Multiple aRISs are deployed between numerous access points (APs) and mobile users to enhance the signals. The active reflective elements (REs) amplify and vary the signal phase. The continuously evolving channel creates a situation, where the channel differs during training and data transmission. We characterize the joint impact of multi-user interference (MUI), pilot contamination (PC), channel aging (CA), and active RIS noise amplification (RNA). The desired signal is enhanced as the reflected signal amplitude and REs per aRIS increase. However, the PC, MUI, CA, and RNA also grow– constraining the SE. Also, the network power consumption increases. An alternating optimization framework based on the weighted minimum mean square error and fractional programming is proposed to optimize the transmit power and reflection coefficients (RCs) with the objective of maximizing the sum SE. It is demonstrated that the number of APs and REs can be reduced by deploying aRISs. The proposed power and RC optimization algorithms considerably improves the sum SE. The trade-off analysis between the total EE and sum SE shows that the envelope of the operating region of the CF mMIMO is expanded by deploying multiple aRISs.