VeXaRIS

Project background

Reconfigurable intelligent surfaces (RIS), massive multiple-input multiple-output (mMIMO) cell-free (CF) technologies and vehicule-to-everything (V2X) are key technologies for next-generation wireless communication, i.e. beyond 5G and 6G. CF mMIMO involves the deployment of numerous access points over a vast geographical area to jointly serve all users.

Project objective and relevance

V2X enables vehicle user equipment (VUE) to communicate with base stations, pedestrians and other vehicles. The high mobility of VUE means frequent handovers and aging channels. CF mMIMO technology is therefore ideal for V2X, since it eliminates the need to change stations.

Approach

To reduce the impact of aging channels, more access points need to be deployed, which increases power consumption. RIS includes numerous passive reflecting elements (RE), which can independently induce phase and amplitude change, with marginal power requirements. RIS can therefore improve spectral/energy efficiency (SE/EE). However, passive RIS suffers from double-fading attenuation that can be detrimental to EUVs.

This proposal aims to analyze the SE/EE of mMIMO CF assisted by active RIS (aRIS) enabled for V2X. Each AP serves the VUEs using several aRIS. We will derive closed-form solutions for the uplink/downlink SE assuming imperfect channel state information, spatially correlated Rician channels and channel aging. For aRIS, while increasing the amplitude of the reflected signal improves the desired signal, it also increases noise. What's more, aRIS requires additional power to amplify the signal, which increases network energy consumption.

Expected results

The trade-off between SE and EE will be analyzed to identify the best operating regions. Increasing the transmission power of APs (or VUEs) may increase SE but saturates at some point due to interference. Consequently, we seek to jointly optimize aRIS power, amplitude and phase with the following objectives: (1) maximize the sum of SE (2) maximize EE. Finally, we propose a federated learning algorithm for CSI prediction to mitigate the effects of channel aging.

 Ce projet a reçu un financement du programme-cadre Horizon Europe (2021-2027) dans le cadre de la convention de subvention n° 101120779