Rate Splitting for MIMO Wireless Networks: A Promising PHY-Layer Strategy for LTE Evolution

Bruno Clerckx; Hamdi Joudeh; Chenxi Hao; Mingbo Dai; Borzoo Rassouli

Source-linked AI summary

Rate Splitting for MIMO Wireless Networks: A Promising PHY-Layer Strategy for LTE Evolution

Bruno Clerckx, Hamdi Joudeh, Chenxi Hao, Mingbo Dai, Borzoo Rassouli

arXiv:1603.01217v1Mar 3, 2016cs.ITcs.NI

TL;DR

MIMO networks are constrained by the difficulty of acquiring accurate CSIT, especially as antenna and access-point counts increase. The paper surveys rate-splitting, which combines common and private messages for imperfect-CSIT operation, and reports benefits in efficiency, reliability, and feedback overhead. It also discusses optimization, deployment scenarios, standardization, and operational challenges.

Problem
Accurate CSIT is increasingly difficult to obtain in dense multiuser MIMO networks, and imperfect CSIT creates multi-user interference that limits performance.
Method
The paper surveys rate-splitting strategies that transmit common and private messages across MU-MIMO, massive MIMO, and multi-cell coordination scenarios.
Results
Rate-splitting is reported to provide benefits in spectral and energy efficiency, reliability, and CSI feedback overhead relative to private-message strategies used in LTE-A.
Takeaways & Limitations
Rate-splitting has the potential to affect LTE Evolution PHY and Lower MAC designs, including transmission, feedback, receivers, scheduling, and reliability.
Takeaways & Limitations
The paper briefly covers selected aspects of MU-MIMO, massive MIMO, and multi-point coordination while identifying broader research and standardization challenges.

Abstract

from arXiv · show

MIMO processing plays a central part towards the recent increase in spectral and energy efficiencies of wireless networks. MIMO has grown beyond the original point-to-point channel and nowadays refers to a diverse range of centralized and distributed deployments. The fundamental bottleneck towards enormous spectral and energy efficiency benefits in multiuser MIMO networks lies in a huge demand for accurate channel state information at the transmitter (CSIT). This has become increasingly difficult to satisfy due to the increasing number of antennas and access points in next generation wireless networks relying on dense heterogeneous networks and transmitters equipped with a large number of antennas. CSIT inaccuracy results in a multi-user interference problem that is the primary bottleneck of MIMO wireless networks. Looking backward, the problem has been to strive to apply techniques designed for perfect CSIT to scenarios with imperfect CSIT. In this paper, we depart from this conventional approach and introduce the readers to a promising strategy based on rate-splitting. Rate-splitting relies on the transmission of common and private messages and is shown to provide significant benefits in terms of spectral and energy efficiencies, reliability and CSI feedback overhead reduction over conventional strategies used in LTE-A and exclusively relying on private message transmissions. Open problems, impact on standard specifications and operational challenges are also discussed.

1. Introduction

Next-generation MIMO deployments face interference as accurate CSIT becomes harder to acquire, motivating rate-splitting designs for imperfect-CSIT networks.

1. Introduction: Dense distributed and co-localized MIMO deployments increase the importance of interference management.Distributed deployments produce dense homogeneous or heterogeneous networks, while co-localized deployments lead to massive MIMO.
1. Introduction: Accurate CSIT is difficult to obtain because pilot reuse, feedback overhead, delays, and RF-chain calibration errors impair channel knowledge.These inaccuracies degrade beamforming, interference nulling, and link adaptation in downlink multi-user MIMO.
1. Introduction: Applying techniques designed for perfect CSIT to imperfect-CSIT scenarios risks widening the gap between theory and practice as antenna density increases.The paper asks whether networks should instead be designed from the outset around imperfect CSIT and its resulting interference.
1. Introduction: Rate-splitting divides messages into common and private parts, superposes a common message over private messages, and lets all receivers decode the common part.The common message is intended for a subset of users but is decodable by all, unlike the private parts decoded only by their corresponding receivers.
1. Introduction: The paper surveys rate-splitting across MU-MIMO, massive MIMO, and multi-cell coordination, highlighting benefits and standardization challenges for LTE evolution.The stated benefits include spectral and energy efficiency, reliability, and reduced CSI feedback overhead.

2. Fundamental of Rate Splitting

Rate-splitting combines common and private transmission to manage interference under imperfect CSIT, with DoF gains illustrated against conventional strategies.

2. Fundamental of Rate Splitting: Rate-splitting divides each user message into common and private parts and packs all common parts into one super common message.The division ratios are design parameters that depend on the system setup.
2. Fundamental of Rate Splitting: The common stream is multicast to all users, decoded before each private stream, and removed through successive interference cancellation.Private streams are decoded after treating them as noise during common-stream decoding.
2. Fundamental of Rate Splitting: Unlike a public multicast message, the RS super common message carries parts of private messages and is decoded by all users for interference mitigation.It is not entirely required by every user even though every user decodes it.
2. Fundamental of Rate Splitting: Under imperfect CSIT, maximum MIMO-BC DoF is maintained when α≥1, whereas practical systems can have α<1.LTE-A quantized CSI is given as an example of a setting where α<1 may occur.
2. Fundamental of Rate Splitting: For single-antenna users with α<1, ZFBF achieves Kα DoF, while RS decoding of common interference adds 1−α DoF.The gain requires power allocation that preserves both common-stream and private-stream DoF.

4. Sum-Rate Enhancement and CSI Feedback Reduction

Finite-SNR analysis examines how rate-splitting improves sum rate and reduces feedback requirements relative to conventional schemes under quantized CSIT.

4. Sum-Rate Enhancement and CSI Feedback Reduction: The sum-rate study optimizes the private-message power fraction ρ as a function of CSIT error, SNR, transmit antennas, and feedback bits.Random Vector Quantization is used for the quantized CSIT in the two-user setting.
4. Sum-Rate Enhancement and CSI Feedback Reduction: Figure 3 compares RS with TDMA, ZFBF, and SU/MU for four transmit antennas, two users, and B=10 or 15 feedback bits.SU/MU switches dynamically between ZFBF and TDMA to maximize sum rate.
4. Sum-Rate Enhancement and CSI Feedback Reduction: RS achieves a higher sum rate than SU/MU, and the rate gap increases as feedback grows from B=10 to B=15.At high SNR, RS continues increasing with DoF 1 while ZFBF saturates under inaccurate CSIT.
4. Sum-Rate Enhancement and CSI Feedback Reduction: At 15 dB and a 6 bps/Hz constant gap target, RS requires 5 fewer feedback bits than ZFBF with RVQ using four transmit antennas.The comparison concerns achieving the same performance relative to ZFBF with perfect CSIT.

5. Transceiver Optimization

RS transceiver optimization balances interference suppression and desired signal power under imperfect CSIT, using non-convex rate formulations that can be reformulated as WMSE problems. Optimized precoders outperform simpler ZFBF-based designs across sum-rate and minimum-rate objectives.

Finite-SNR precoder design must balance nulling undesired interference against maximizing desired received power for metrics such as rate, SINR, and MSE.
RS optimization divides each user’s achievable rate into common and private components, making the message-splitting ratio a design variable.
Non-convex sum-rate objectives can be reformulated as equivalent WMSE problems and solved by alternating optimization, although global optimality is not guaranteed.
Optimized precoders outperform simpler ZFBF-based designs for both average sum-rate and minimum-average-rate objectives with i.i.d. Gaussian CSIT errors at α=0 and 0.6.

6. Massive MIMO

Massive MIMO makes full-dimensional CSIT costly or unreliable, motivating hierarchical rate-splitting. HRS uses correlation-based grouping and layered common messages to address inter-group and intra-group interference.

Full-dimensional channel estimation in massive MIMO requires unaffordable FDD feedback or suffers pilot contamination and antenna/RF miscalibration in TDD.
HRS exploits transmit correlation matrices and uses two kinds of common messages to mitigate the rate degradation caused by decoding one common message across many users.
Users are grouped by transmit-correlation similarity, while outer and inner RS layers address inter-group and intra-group interference, respectively.
HRS outperforms RS and conventional approaches in the illustrated massive-MIMO scenario with M=100 and SNR=30dB.

7. Multi-Cell Coordination

Multi-cell coordination is limited by imperfect CSIT and variable reliability, so rate-splitting extends from two-cell coordination to topology-aware multi-layer designs. In the two-cell case, RS improves the sum DoF over ZFBF under symmetric cross-link CSIT quality.

CoMP addresses inter-cell interference in LTE-A, but imperfect CSIT among coordinated transmitters affects throughput and reported performance varies substantially.
In the two-cell scenario, ZFBF achieves sum DoF 2α, whereas RS increases it to 1 + α by combining reduced-power private streams with one common message.
Three-cell interference can require multiple common messages because a single common message may not properly address interference from two distinct cross links.
TRS uses CSIT-quality topology to organize a multi-layer rate-splitting structure, including grouping users with identical intra-group qualities.

8. Rate-Splitting in LTE Evolution

(H/T)RS unifies conventional and rate-splitting transmission modes, while LTE Evolution standardization requires signaling message types and demodulation information.

8. Rate-Splitting in LTE Evolution: (H/T)RS includes conventional SU/MU-MIMO and CoMP when common-message power is zero.This supports mode switching among SU, conventional MU, and RS according to SNR and CSIT quality.
8. Rate-Splitting in LTE Evolution: LTE Evolution needs a new DCI transmission-mode indicator carrying the information required for demodulation.Receivers must be informed about common/private message types, message counts, and modulation and coding schemes.

9. Conclusions and Future Challenges

The paper presents RS as a strategy for imperfect-CSIT LTE Evolution and identifies broad potential benefits alongside extensive future research and standardization challenges.

9. Conclusions and Future Challenges: RS uses common and private messages for realistic imperfect-CSIT scenarios, unlike LTE-A’s private-message design.The paper highlights spectral efficiency, energy efficiency, reliability, and CSI feedback overhead reduction benefits over conventional LTE-A strategies.
9. Conclusions and Future Challenges: RS could affect PHY and Lower MAC Layer design across transmission, feedback, receiver, scheduling, waveform, and efficiency considerations.The paper frames these areas as research problems for academia and standard-specification issues for industry.
9. Conclusions and Future Challenges: Future challenges span millimeter-wave operation, distributed-antenna coordination, interference alignment, network MIMO, relaying, multicast-unicast superposition, and proactive caching.