Authors: Volker Jungnickel (Fraunhofer Heinrich Hertz Institute), Konstantinos Manolakis (Fraunhofer Heinrich Hertz Institute), Wolfgang Zirwas (Nokia Solutions and Networks), Berthold Panzner (Nokia Solutions and Networks), Volker Braun (Alcatel Lucent Bell Labs), Moritz Lossow (Deutsche Telekom AG, Innovation Labs), Mikael Sternad (Uppsala University), Rikke Apelfröjd (Uppsala University), and Tommy Svensson (Chalmers University of Technology)
Title: “The Role of Small Cells, Coordinated Multipoint, and Massive MIMO in 5G”
Publication: May, 2014 IEEE Communications Magazine
The authors focus on three key technologies for achieving higher spectral efficiency in 5G: interference mitigation (through coordinated multipoint), network “densification” (through small cells), and massive MIMO. They outline an architecture where these three technologies work together, as outlined below.
Joint transmission coordinated multipoint (JT CoMP) involves simultaneous downlink transmissions from a small number of base stations to an end user. The authors identify recent demonstrations of JT CoMP validating its technical feasibility, but which also identify the need for refinements to improve performance. The authors identify two key areas for improvement as i) clustering and user selection, and ii) feedback compression and channel prediction.
First, JT CoMP involves both identifying which user(s) should receive joint transmissions, as well as which collection of base stations is best-suited to serving the selected users. Early results identified this optimal selection problem as NP-hard, and as such efficient heuristics are suitable. The authors propose a “cover-shift” concept for forming overlapping cooperation areas (CAs), particularly for the cell-edge users who are particularly vulnerable to inter-cell interference. The authors also suggest a “successive user grouping” heuristic for user selection wherein users are successively added to a JT CoMP group provided the addition of the user does not reduce the CoMP performance gain of the group.
Second, the performance gains of JT CoMP are known to be particularly sensitive to channel estimation errors arising from feedback quantization and delay. To address this concern, the authors have investigated feedback compression and channel prediction techniques. Feedback compression is achieved by combining user clustering, tap selection, and adaptive channel quantization; their implementation reduced the channel state information (CSI) feedback overhead by a factor of 15 relative to the LTE reference case. Furthermore, the authors have implemented a recently developed channel prediction incorporating Doppler effects which have demonstrated improvements in terms of mean squared error of channel predictions relative to conventional techniques.
In the last two main sections of the paper, the authors outline a proposed integrated 5G design incorporating small cells, JT CoMP, and massive MIMO, and present some initial performance results. The synergy between JT CoMP and massive MIMO is due to the fact that massive MIMO helps to “localize” the interference, which facilitates improvements in coordinated transmissions, as the participating base stations have a reduced need for coordination. Their initial results are for the combined use of small cell architecture with JT CoMP, and are based on ray tracing simulations in the Munich area. By extrapolating their results to incorporate anticipated massive MIMO antenna designs, the authors predict this architecture capable of supporting 1000 times what can be supported with LTE.