Institutional Repository
Technical University of Crete
EN  |  EL



My Space

Practical channel estimation for reconfigurable reflection surfaces in next generation wireless networks

Kotridis Georgios

Full record

Year 2021
Type of Item Diploma Work
Bibliographic Citation Georgios Kotridis, "Practical channel estimation for reconfigurable reflection surfaces in next generation wireless networks", Diploma Work, School of Electrical and Computer Engineering, Technical University of Crete, Chania, Greece, 2021
Appears in Collections


After the introduction of the Multiple Input Multiple Output and mmWave technologies to the communication world, great research interest has been recently attracted to find new ways of meeting the need for higher data rates that are both energy efficient and have low hardware cost. Reconfigurable Intelligent Surface (RIS) is a hardware technology composed of passive, software-controlled metamaterials with reconfigurable scattering properties, meaning that the right adjustments can lead to constructive addition of EM waves at the receiver’s end. Due to the passive nature of the elements there can be no signal processing of the incoming signals between the transmitter and the receiver, leading to challenging channel estimation. This work first studies prior art in the literature regarding channel estimation for MIMO systems using the zero-mean error Linear Minimum Mean Square Estimator (LMMSE), where the estimator is an affine transformation of the received signal. It offers in-depth derivation of both the estimator and the MSE formulas, which can be applied to any MIMO system, including the RIS case for Rayleigh and Rician fading, showed first theoretically and then corroborated by simulation results. We then use the estimation of all the channel coefficients as input-data to a novel algorithm, which computes the optimal element-configuration in O(Mlog(M)) time, where M is the number of RIS elements. It is noted that the optimal configuration could not be found with exhaustive search, when M is in the order of thousands, due to the inherent exponential complexity. Simulations indicate that after the algorithm’s application, the average power improvement reaches 12 dB gain for M = 6162, while the impact of channel estimation error for the same value of M leads to a gain loss in the order of 3 dB.

Available Files