PaperlayerView exact revision on arXiv

Source-linked AI summary

MmWave Massive MIMO Based Wireless Backhaul for 5G Ultra-Dense Network

Zhen Gao, Linglong Dai, De Mi, Zhaocheng Wang, Muhammad Ali Imran, Muhammad Zeeshan Shakir

arXiv:1508.03940v3cs.IT

TL;DR

The paper addresses how to provide reliable, Gigahertz-bandwidth, cost-effective backhaul for 5G UDN. It proposes DPSN-based hybrid precoding/combining with compressive-sensing channel estimation, leveraging channel low rank to reduce transceiver cost and complexity with negligible performance loss. The approach supports multiple small-cell BSs and multiple streams per small-cell BS while the paper also examines deployment benefits and implementation challenges.

  • Problem

    5G UDN needs reliable, Gigahertz-bandwidth, cost-effective backhaul, while mmWave massive MIMO introduces transceiver and channel-state-information challenges.

  • Method

    The paper proposes DPSN-based hybrid precoding/combining with associated compressive-sensing channel estimation, leveraging the low-rank mmWave massive MIMO channel matrix.

  • Results

    The proposed scheme supports simultaneous service to multiple small-cell BSs with multiple streams for each small-cell BS and incurs negligible capacity loss versus optimal full-digital processing.

  • Takeaways & Limitations

    MmWave massive MIMO provides a viable wireless-backhaul approach for UDN with flexible topology and scheduling while reducing required transceiver cost and complexity.

Abstract

from arXiv · show

Ultra-dense network (UDN) has been considered as a promising candidate for future 5G network to meet the explosive data demand. To realize UDN, a reliable, Gigahertz bandwidth, and cost-effective backhaul connecting ultra-dense small-cell base stations (BSs) and macro-cell BS is prerequisite. Millimeter-wave (mmWave) can provide the potential Gbps traffic for wireless backhaul. Moreover, mmWave can be easily integrated with massive MIMO for the improved link reliability. In this article, we discuss the feasibility of mmWave massive MIMO based wireless backhaul for 5G UDN, and the benefits and challenges are also addressed. Especially, we propose a digitally-controlled phase-shifter network (DPSN) based hybrid precoding/combining scheme for mmWave massive MIMO, whereby the low-rank property of mmWave massive MIMO channel matrix is leveraged to reduce the required cost and complexity of transceiver with a negligible performance loss. One key feature of the proposed scheme is that the macro-cell BS can simultaneously support multiple small-cell BSs with multiple streams for each smallcell BS, which is essentially different from conventional hybrid precoding/combining schemes typically limited to single-user MIMO with multiple streams or multi-user MIMO with single stream for each user. Based on the proposed scheme, we further explore the fundamental issues of developing mmWave massive MIMO for wireless backhaul, and the associated challenges, insight, and prospect to enable the mmWave massive MIMO based wireless backhaul for 5G UDN are discussed.

I. INTRODUCTION

The paper motivates mmWave massive MIMO wireless backhaul as a reliable, high-bandwidth, cost-effective foundation for 5G UDN and introduces DPSN-based hybrid processing to address implementation challenges.

  • I. INTRODUCTION: UDN requires reliable, cost-effective backhaul with 1–10 GHz bandwidth connecting macro-cell and small-cell BSs.Wireless mmWave backhaul is presented as an alternative to optical fiber where deployment and installation constraints limit economy.
  • I. INTRODUCTION: Underutilized mmWave spectrum can provide potential Gigahertz transmission bandwidth for UDN backhaul.This contrasts with the scarce microwave spectrum used in conventional cellular networks.
  • I. INTRODUCTION: Massive antennas at mmWave frequencies improve signal directivity, reduce co-channel interference, and enhance backhaul link reliability.The small mmWave wavelength makes large antenna arrays practical.
  • I. INTRODUCTION: The article examines the feasibility, benefits, differences, challenges, and research directions of mmWave massive MIMO backhaul for UDN.It specifically stresses the spatial sparsity of mmWave massive MIMO channels.
  • I. INTRODUCTION: The proposed DPSN-based hybrid precoding/combining scheme is paired with compressive-sensing channel estimation for cost-effective mmWave massive MIMO backhaul.The scheme is designed to address the transceiver cost and complexity associated with mmWave massive MIMO.

II. FEASIBILITY AND CHALLENGES OF MMWAVE MASSIVE MIMO FOR WIRELESS BACKHAUL IN 5G UDN

MmWave massive MIMO suits UDN backhaul because it offers capacity, compact deployment, and reliable directional transmission, while introducing transceiver, channel-estimation, and channel-state-information challenges.

  • II. FEASIBILITY AND CHALLENGES OF MMWAVE MASSIVE MIMO FOR WIRELESS BACKHAUL IN 5G UDN: UDN backhaul must provide large bandwidth and reliable transmission while controlling power efficiency and deployment cost.Small cells are densely deployed in traffic hotspots to offload macro-cell traffic.
  • II. FEASIBILITY AND CHALLENGES OF MMWAVE MASSIVE MIMO FOR WIRELESS BACKHAUL IN 5G UDN: More than one Gbps backhaul capacity can be supported over a 250 MHz E-band channel.The paper identifies underutilized V-band and E-band spectrum as sources of potential Gigahertz bandwidth.
  • II. FEASIBILITY AND CHALLENGES OF MMWAVE MASSIVE MIMO FOR WIRELESS BACKHAUL IN 5G UDN: MmWave’s high path loss can support improved frequency reuse and reduced intercell interference in UDN deployments.The paper describes E-band transmission over several kilometers and V-band transmission over about 500–700 m under attenuation constraints.
  • II. FEASIBILITY AND CHALLENGES OF MMWAVE MASSIVE MIMO FOR WIRELESS BACKHAUL IN 5G UDN: Small mmWave wavelengths enable massive antennas, directional transmission, path-loss compensation, compact equipment, and deployment at low-cost sites.Examples include light poles, building walls, and bus stations.
  • II. FEASIBILITY AND CHALLENGES OF MMWAVE MASSIVE MIMO FOR WIRELESS BACKHAUL IN 5G UDN: MmWave massive MIMO inherits flexible beamforming, spatial multiplexing, and diversity, improving backhaul reliability and enabling flexible network architectures.Its implementation differs from microwave massive MIMO used in radio access networks.
  • II. FEASIBILITY AND CHALLENGES OF MMWAVE MASSIVE MIMO FOR WIRELESS BACKHAUL IN 5G UDN: High-cost mmWave transceiver components make massive low-cost antennas with limited expensive baseband chains appealing but challenge conventional precoding and combining.The cited components include high-speed ADCs and DACs, synthesizers, and mixers.
  • II. FEASIBILITY AND CHALLENGES OF MMWAVE MASSIVE MIMO FOR WIRELESS BACKHAUL IN 5G UDN: Large mmWave antenna arrays make channel estimation more difficult and make synchronization and RF-chain calibration errors nontrivial even with TDD reciprocity.These issues constrain practical exploitation of channel reciprocity.
  • II. FEASIBILITY AND CHALLENGES OF MMWAVE MASSIVE MIMO FOR WIRELESS BACKHAUL IN 5G UDN: MmWave backhaul requires CSIR feedback to small-cell BSs because uplink precoding and downlink combining support multiple streams and directional transmission.This requirement extends beyond the CSIT-only assumption typical for single-antenna microwave users.

III. MMWAVE CHANNEL CHARACTERISTICS

MmWave massive MIMO channels are spatially and angularly sparse, giving them a low-rank structure despite large antenna dimensions. This limits effective independent streams and permits fewer baseband chains with negligible performance loss under favorable conditions.

  • Low-rank property: With fewer baseband chains than antennas, performance loss depends largely on the propagation conditions of mmWave channels.The reduced-chain architecture is most suitable when channel sparsity supports a small effective rank.
  • Spatial/angular sparsity: MmWave channels typically contain only 3∼5 dominant multipaths because of high path loss for non-line-of-sight signals.The channel model describes paths through complex gains and azimuth AoD/AoA parameters.
  • Low-rank property: Small path count combined with dozens or hundreds of antennas gives mmWave massive MIMO channel matrices an obvious low-rank property.The effective rank is much smaller than the full antenna dimensions.
  • Low-rank property: The number of effective independent streams is small, so capacity exhibits a ceiling as the number of baseband chains increases.This motivates matching the number of baseband chains to the channel’s effective rank.

IV. KEY ISSUES OF DESIGNING MMWAVE MASSIVE MIMO FOR 5G UDN BACKHAUL

UDN backhaul must support multiple small-cell BSs and multiple streams per BS, challenging conventional precoding architectures. The proposed DPSN-based hybrid design exploits channel low rank to reduce transceiver complexity while retaining multi-user, multi-stream transmission.

  • Design requirements: Reliable point-to-multiple-points backhaul requires flexible beamforming and spatial multiplexing for multiple small-cell BSs and streams per BS.This requirement exceeds the single-user or single-stream-per-user assumptions of conventional schemes.
  • Design requirements: Conventional analog precoding supports only SU-MIMO with one stream, while full digital precoding requires one RF chain per antenna.Conventional hybrid schemes motivate a lower-cost architecture for mmWave backhaul.
  • Proposed DPSN design: The DPSN-based hybrid scheme supports multi-user and multi-stream transmission while reducing transceiver cost and complexity.It combines analog and digital processing while exploiting the effective low-rank channel dimensions.
  • Proposed DPSN design: The channel for each small-cell BS is approximated through SVD using an effective rank R_k, separating dominant singular components from residual components.The approximation follows from the low-rank property of the mmWave massive MIMO channel.
  • Proposed DPSN design: Analog and digital precoders are iteratively obtained to approximate the target precoder while enforcing constant-modulus analog elements.The update uses phase matching, a pseudoinverse-based digital step, and repeated refinement until convergence.
  • Proposed DPSN design: The resulting equivalent channel can be diagonalized to realize simultaneous multi-user and multi-stream transmission.The uplink uses an analogous precoding and combining procedure.

B. CSI Acquisition for MmWave Massive MIMO

Reliable, low-overhead CSI acquisition is necessary to implement the DPSN-based hybrid precoding and combining scheme.

  • CSI acquisition: CSI acquisition must provide reliable channel information with low overhead for the proposed DPSN-based hybrid scheme.The requirement is identified as a separate implementation challenge.

1) Challenging Channel Estimation for MmWave Massive MIMO:

Channel estimation is difficult because mmWave massive MIMO combines many antennas with few baseband chains, phase-shifter networks, RF imperfections, and highly directional links. Reliable estimation therefore requires additional signal power and partial transmitter-side channel information.

  • Estimation challenges: Massive antenna arrays can create prohibitively high channel-estimation overhead, while RF-chain calibration and synchronization errors remain significant in TDD.Using fewer baseband chains also reduces the effective dimensions available for channel estimation.
  • Estimation challenges: Baseband channel estimation must account for phase-shifter networks at both macro-cell and small-cell BSs, increasing complexity.The hybrid architecture changes the effective observation process at both ends of the link.
  • Estimation challenges: Strong mmWave signal directivity makes reliable channel estimation dependent on sufficient received power and at least partial CSIT for transmitter beamforming.Beam alignment is needed to match the transmitter pattern to the mmWave MIMO channel.

2) Overview of Existing Channel Estimation Schemes:

IEEE 802.11ad uses three beamforming-training phases, while IEEE 802.15.3c uses codebooks for indoor scenarios with few antennas.

  • IEEE 802.11ad training uses sector-level sweep, beam refinement, and beam tracking to compensate 60 GHz path loss.The phases select sectors, refine beams, and adjust beamforming during transmission.

3) Proposed CS-Based Channel Estimation for MmWave Massive MIMO:

The proposed channel-estimation approach exploits long coherence time and low-rank mmWave channels through compressed sensing, using coarse beam acquisition followed by parameter estimation at both ends.

  • Long channel coherence in fixed backhaul links reduces the frequency of required channel estimation relative to RAN channels.
  • Low-rank mmWave channel matrices have small effective DoF, enabling channel reconstruction from significantly reduced measurements using compressed sensing.
  • Phase 1 obtains partial CSIT by sweeping predefined transmit and receive beamforming patterns, then feeds back several optimal pattern indices.
  • Phase 2 estimates AoA and path gains at each small-cell BS using FRI theory and analog compressed sensing with feedback-guided beamforming.
  • Phase 3 estimates AoD and path gains at the macro-cell BS, using FRI-based super-resolution to distinguish paths with small angular differences.
  • Further work includes optimizing coarse beam patterns, designing training signals, and developing low-complexity, high-accuracy compressed-sensing channel estimation.

4) Other Issues of Channel Estimation:

Future channel-estimation work must address feedback, beam-pattern design, channel tracking, and extensions from ULA to three-dimensional MIMO.

  • Future designs should optimize coarse beamforming, training signals, channel feedback, and dynamic tracking for blockage or slow channel changes.A limited-resource microwave control link may feed back typically 3–5 AoA/AoD parameters.
  • For urban UDN, extending the ULA scheme to 3D MIMO can exploit both azimuth and elevation for improved backhaul performance.
  • 3D MIMO increases computational concerns for SVD and waterfilling, motivating low-complexity schemes based on angular sparsity, geometry, and sub-optimal power allocation.

D. Sampling with Low-Resolution ADC

The performance study compares DPSN hybrid precoding/combining with full digital processing under ideal and compressed-sensing channel information, showing near-full-digital capacity with fewer baseband chains.

  • The simulations use 60 GHz ULA links with K = 4, while channel models vary path count, angles, and Rician LOS-to-NLOS power ratio.
  • The proposed hybrid scheme incurs negligible capacity loss versus optimal full digital processing despite using far fewer baseband chains.The low-rank channel structure produces a capacity ceiling when the number of baseband chains is sufficiently large.
  • With more measurements, compressed-sensing channel estimation approaches the capacity achieved with ideal CSI.
  • Waterfilling power allocation outperforms equal-power allocation, supporting its consideration in practical backhaul design.

VI. BENEFITS AND OPPORTUNITIES OF MMWAVE MASSIVE MIMO BASED BACKHAUL NETWORK

The proposed mmWave massive MIMO backhaul supports multi-small-cell connectivity through P2MP topology and flexible spatial scheduling. TDD offers spectrum and traffic-adaptation benefits, while practical deployment still requires interference management and improved regulation.

  • Point-to-Multiple-Points Backhaul: The proposed mmWave massive MIMO backhaul enables a macro-cell BS to simultaneously support multiple small-cell BSs, providing a viable mmWave P2MP approach.P2MP is associated with lower total cost of ownership than P2P in conventional backhaul networks.
  • Beam Division Multiplex (BDM) Based Scheduling: BDM scheduling uses flexible spatial multiplexing and hybrid beamforming to reduce reliance on separate time slots for different backhaul links.The macro-cell BS can combine BDM and TDM according to backhaul load, assigning more beam resources to heavy-load or non-LOS links.
  • TDD is Suitable for MmWave Backhaul Network: TDD shares one frequency band for uplink and downlink, allowing flexible slot allocation for asymmetric backhaul traffic.The passage contrasts this with FDD, which uses different uplink and downlink bands and may face country-specific regulation.
  • TDD is Suitable for MmWave Backhaul Network: Practical TDD mmWave massive MIMO backhaul requires adaptive interference management, automated configuration, and improved licensed E-band spectrum regulation.These needs address mutual interference across operators and plug-and-play deployment, especially in unlicensed V-band.
  • Conclusions: By leveraging the low-rank channel property, the proposed DPSN hybrid precoding/combining scheme supports multiple small-cell BSs with multiple streams each.The article presents this scheme as an approach toward P2MP topology and BDM scheduling for mmWave backhaul.
Loading 1508.03940v3…