Abstract
In this letter, we revisit hybrid analog–digital precoding systems with emphasis on the modeling of their radio-frequency (RF) losses, to realistically evaluate their benefits in 5G system implementations. We focus on fully-connected analog beamforming networks (FC-ABFNs) and on discrete Fourier transform implementations, and decompose these as a bank of commonly used RF components. We then model their losses based on their S-parameters. Our results reveal that the performance and energy efficiency of hybrid precoding systems are severely affected once these, commonly ignored, losses are considered in the overall design. In this context, we also show that hybrid precoder designs similar to Butler matrices are capable of providing better performances than FC-ABFN for systems with a large number of RF chains.
Highlights
HYBRID PRECODING UNDERLet us consider a base stations (BS) comprised of N antennas transmitting towards K ≤ N single-antenna users
A variety of ABFN have been recently proposed [2]–[4]. They generally disregard the practical implications of signal processing in the RF domain such as the additional power losses introduced by the ABFN
The conclusions derived in the following can be applicable to a vast number of hybrid precoding designs, in the following we focus on the joint spatial division and multiplexing (JSDM) of [4], since it admits both fullyconnected and DFT-based designs
Summary
Let us consider a BS comprised of N antennas transmitting towards K ≤ N single-antenna users. The adaptive nature of FBB and the data symbols produces phase and amplitude mismatches in the signals at the input of the power combiners, introducing a loss in the signal combining process due to their non-coherent addition [9] - an aspect not often considered in the related literature We refer to this mismatch leading to loss among fixed connections in FRF as dynamic power loss [6], [7], and we remark that it arises even for ideal analog hardware components. We propose to consider 4-port hybrid directional couplers This is because, in contrast with Wilkinson combiners, these components ideally preserve the power introduced at their input, even when the input signals have different phases and amplitudes [6], [7]. For an illustrative frequency of 2.6 GHz and arbitrary FBB, we have observed that the dynamic signal-dependent loss is approximately zero and that the static loss approaches 2.8 dB
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
