Evaluation of Channel Estimation Algorithms Using Practically Measured Channels in FDD Massive MIMO

Abstract

An important problem for massive multiple-input multiple-output (MIMO) systems operating with frequency-division duplexing (FDD) is to accurately estimate the channel response with low pilot signal overhead. Most existing algorithms for efficient channel estimation are based on compressive sensing (CS) and assume sparse structure of the channel vector. Relying on it, they try to minimize estimation error and reduce the number of required pilot signals. Utilizing real-world channel responses, we evaluate the performance of 11 state-of-the-art channel estimation algorithms for FDD massive MIMO systems. Results from simulation experiments with channel measurements for carrier frequency in the 2.4 GHz and 5 GHz bands for three environments and two levels of mobility are presented. Channel structures of theoretical and practically measured channels are compared and it is shown that the latter does not follow a specific sparse structure which leads to a significant increase in estimation errors according to our results. A comprehensive analysis of estimation quality and its dependence on signal-to-noise ratio (SNR) and number of pilot signals is provided. The results demonstrate that some algorithms perform well when applied to practical channels while others do not provide confident results. The effects of pilot matrix choice and angular domain channel representation are also studied and evaluated.

Authors

  • Nikolay Dandanov
  • Krasimir Tonchev
  • Vladimir Poulkov
  • Pavlina Koleva

Venue

International Conference on Future Access Enablers of Ubiquitous and Intelligent Infrastructures, 2019.

Links

https://link.springer.com/chapter/10.1007/978-3-030-23976-3_21

Categories

, , ,