Abstract
We propose a low-complexity approach for the downlink of physically constrained massive multiple-input multiple-output (MIMO) systems with user mobility. We examine a channel state information (CSI) acquisition strategy that exploits both the spatial and temporal correlations among the channels of adjacent base station antennas. The proposed strategy solely collects CSI for a subset of antennas and time frames. Then full CSI is approximated using the CSI of adjacent antennas and previous frames. This critically reduces the CSI acquisition complexity while sacrificing the CSI quality and, hence, introduces a scalable performance-complexity tradeoff. The numerical results demonstrate that, for practical mobile speeds, the proposed scheme reduces the computational complexity and enhances the energy efficiency of massive MIMO base stations against systems with complete CSI, while approximately preserving performance.
Highlights
Massive multiple-input multiple-output (MIMO) has received great attention thanks to its potential to address the increasing data rate demands via serving a significant number of users simultaneously [1]
Temporal channel correlation and the detected data symbols can be leveraged in time division duplex (TDD) massive MIMO systems to address pilot contamination [5], while the correlation between antennas is employed to optimize the channel state information (CSI) acquisition stage in frequency division duplex (FDD) systems [6]
1While the proposed technique could be applied to massive MIMO systems operating in FDD, in the following we focus on the more practical TDD systems [1]
Summary
While massive MIMO promotes the use of linear precoding techniques such as zero forcing, the colossal number of antennas still imposes significant computational complexity requirements [2], [3] Since these techniques rely on the availability of instantaneous CSI, a number of channel estimation schemes have been developed to enhance its acquisition [1], [3], [4]. For those frames without an explicit CSI acquisition phase, the channel coefficients are estimated from the CSI previously available by exploiting the channel’s temporal correlation related to user mobility This extends the results of [11] where only spatial correlation is exploited in scenarios without user mobility. Our analysis and simulations show an overall favourable trade-off for the proposed strategy with partial CSI
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have