Abstract
In frequency division duplex (FDD) multiple-input multiple-output (MIMO) wireless communications, limited channel state information (CSI) feedback is a central tool to support advanced single- and multi-user MIMO beamforming/precoding. To achieve a given CSI quality, the CSI quantization codebook size has to grow exponentially with the number of antennas, leading to quantization complexity, as well as, feedback overhead issues for larger MIMO systems. We have recently proposed a multi-stage recursive Grassmannian quantizer that enables a significant complexity reduction of CSI quantization. In this paper, we show that this recursive quantizer can effectively be combined with deep learning classification to further reduce the complexity, and that it can exploit temporal channel correlations to reduce the CSI feedback overhead.
Highlights
L IMITED channel state information (CSI) feedback is a well-established technique for supporting efficient multiple-input multiple-output (MIMO) transmissions in frequency division duplex (FDD) systems [1]–[4]
In case of memoryless quantization of isotropic channels, such as, independent and identically distributed (i.i.d.) Rayleigh fading channels, it is known that maximally spaced subspace packings achieve optimal quantization performance in terms of subspace chordal distance; such packings are difficult to construct for larger MIMO systems and codebook sizes [9]–[11]
We focus on Grassmannian CSI quantization at the receiver, in order to provide CSI feedback to the transmitter
Summary
L IMITED channel state information (CSI) feedback is a well-established technique for supporting efficient multiple-input multiple-output (MIMO) transmissions in FDD systems [1]–[4]. When adopting Grassmannian quantization in larger-scale MIMO systems and/or for high resolution quantization, one faces two main challenges: 1) quantization complexity and 2) feedback overhead The former issue can effectively be tackled, if the channel exhibits structure that can be exploited for quantization; e.g., in the millimeter wave band, the channel is often assumed to be sparse, which allows for efficient parametric CSI quantization by sparse decomposition [12]–[14]. When the channel exhibits temporal correlation, quantizers with memory, such as, differential quantizers or techniques based on recurrent neural networks, can provide significantly better performance than memoryless approaches [18]–[25] They mostly require adaptation of the quantization codebook on the fly or online neural network learning, which can be prohibitive in terms of complexity. The zeroth-order Bessel function of the first kind is J0(·)
Sign-in/Register to view highlights & summary 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 callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
