PaperlayerView exact revision on arXiv

Source-linked AI summary

Heterogeneous Cloud Radio Access Networks: A New Perspective for Enhancing Spectral and Energy Efficiencies

Mugen Peng, Yuan Li, Jiamo Jiang, Jian Li, Chonggang Wang

arXiv:1410.3028v1cs.NI

TL;DR

H-CRANs address severe inter-tier interference and limited cooperation in heterogeneous networks by incorporating cloud computing into HetNets. The paper surveys H-CRAN architectures, efficiencies, techniques, advances, and challenges.

  • Problem

    Severe inter-tier interference and limited cooperative gains constrain heterogeneous networks, while non-ideal backhaul can limit cooperation.

  • Method

    The article surveys H-CRAN architectures, performance analysis, recent advancements, and cloud-computing-based techniques including cooperative resource management and coordinated transmission.

  • Results

    H-CRANs are presented as a promising paradigm combining cloud computing and HetNets to achieve high spectral and energy efficiencies.

  • Takeaways & Limitations

    The surveyed H-CRAN solutions provide advances in theories and technologies for next-generation wireless communication systems.

  • Takeaways & Limitations

    A closed-form spectral-efficiency expression with stochastic geometry for C-RANs remains unavailable.

Abstract

from arXiv · show

To mitigate the severe inter-tier interference and enhance limited cooperative gains resulting from the constrained and non-ideal transmissions between adjacent base stations in heterogeneous networks (HetNets), heterogeneous cloud radio access networks (H-CRANs) are proposed as cost-efficient potential solutions through incorporating the cloud computing into HetNets. In this article, state-of-the-art research achievements and challenges on H-CRANs are surveyed. In particular, we discuss issues of system architectures, spectral and energy efficiency performances, and promising key techniques. A great emphasis is given towards promising key techniques in H-CRANs to improve both spectral and energy efficiencies, including cloud computing based coordinated multi-point transmission and reception, large-scale cooperative multiple antenna, cloud computing based cooperative radio resource management, and cloud computing based self-organizing network in the cloud converging scenarios. The major challenges and open issues in terms of theoretical performance with stochastic geometry, fronthaul constrained resource allocation, and standard development that may block the promotion of H-CRANs are discussed as well.

I. INTRODUCTION

H-CRANs are proposed to address severe interference, constrained cooperation, and energy-efficiency challenges in dense HetNets by combining cloud computing with heterogeneous radio access networks.

  • Dense LPN deployment increases capacity in hotspots but can cause severe inter-tier interference that restricts HetNet gains and commercial development.
  • CoMP can mitigate interference, but its gains depend heavily on backhaul constraints and may degrade as LPN density increases.
  • About 20 percent average uplink CoMP spectral-efficiency gain was reported in downtown Dresden field trials with non-ideal backhaul and distributed processing at base stations.
  • H-CRANs combine HetNets with cloud computing to alleviate inter-tier interference, improve cooperative processing, and enhance both spectral and energy efficiencies.
  • The proposed architecture strengthens HPNs with massive MIMO and connects simplified LPNs to a signal-processing cloud over high-speed optical fibers.
  • The article surveys H-CRAN architectures, SE/EE performance, cloud-based PHY processing, cooperative RRM, self-organization, challenges, and standards.

II. SYSTEM ARCHITECTURE AND PERFORMANCE ANALYSIS OF H-CRANS

H-CRANs integrate HPNs into C-RANs to retain coverage and control signaling while using cloud capabilities to address fronthaul, interference, and energy challenges.

  • HetNets provide coverage and capacity but face constrained backhaul for CoMP and low energy efficiency from ultra-dense LPN deployment.
  • C-RANs centralize processing and can reduce capital and operating expenditures while providing high transmission bit rates and energy efficiency.
  • H-CRANs incorporate HPNs into C-RANs, combining HetNet coverage and capacity with cloud computing capabilities.

A. System Architecture of H-CRANs

H-CRAN architecture separates coverage and control functions at HPNs from high-rate radio access and centralized cooperative processing through RRHs and the BBU pool.

  • H-CRANs connect many low-complexity RRHs to a centralized BBU pool, while HPNs interface with the pool to mitigate cross-tier interference.
  • Centralized cloud processing enables large-scale cooperative processing, suppresses inter-RRH interference, and supports diversity and multiplexing gains.
  • HPNs deliver control signaling and system broadcasts, decoupling control from data symbols and simplifying RRH fronthaul capacity and delay constraints.
  • RRHs can enter sleep mode under low traffic and support burst traffic, while adaptive signaling reduces radio connection and release overhead.
  • Massive MIMO at HPNs and diverse PHY technologies at RRHs provide potential coverage and transmission-rate enhancements.
  • Cloud-based cooperative RRM through the X2 interface mitigates cross-tier interference between HPNs and RRHs.

B. Spectral and Energy Efficiencies Performances

H-CRAN performance depends on balancing centralized cooperation, RRH activation, association size, fronthaul consumption, and mobility-specific service roles.

  • Approximately 60 percent energy-saving opportunities are reported when RRHs switch off during no-traffic periods, compared with non-sleep operation.
  • H-CRANs improve SE and reduce circuit energy consumption by assigning coverage and control to HPNs while RRHs provide high-rate packet traffic.
  • EE decreases as cell-edge UE count increases, while two-tier HetNets outperform one-tier HPNs and two-tier H-CRANs outperform one-tier C-RANs in the described comparisons.
  • Larger user-centric clusters can improve SE but increase fronthaul consumption, requiring cluster-size choices that balance cooperative gains and overhead.
  • Associating two nearest RRHs provides a significant capacity gain over one, while gains among four, eight, and infinitely many RRHs are not large.
  • H-CRANs improve mobility performance relative to C-RANs by serving high-mobility UEs through reliable HPN connections and low-mobility UEs through RRHs.

III. PROMISING KEY H-CRAN TECHNOLOGIES

H-CRANs combine cloud computing with heterogeneous networks to coordinate transmission, antenna processing, radio resources, and self-organization for improved spectral and energy efficiency.

  • CC-CoMP: CC-CoMP coordinates RRHs and HPNs to cancel interference and enable collaboration.It extends traditional 4G CoMP through cloud-based processing.
  • LS-CMA: LS-CMA equips HPNs with large-scale antenna arrays to provide diversity and multiplexing gains.The technique uses large-scale signal processing in HPNs.
  • CC-CRRM: CC-CRRM shares and virtualizes radio resources among RRHs while coordinating cross-interference between RRHs and HPNs.Its scope includes cooperative resource control across heterogeneous tiers.
  • Challenges: Compared with HetNets and C-RANs, H-CRANs incur greater complexity and cost in network planning and maintenance.This is a stated deployment and operation challenge.
  • CC-SON: CC-SON is presented as indispensable for increasing network intelligence and reducing human operating costs.It supports cloud-based self-organization in H-CRANs.

A. Cloud Computing based Coordinated Multi-Point (CC-CoMP)

CC-CoMP uses centralized cloud processing to coordinate transmissions across H-CRAN tiers, alleviating traditional CoMP challenges while introducing scale, CSI, and computational trade-offs.

  • Scenarios: Intra-tier CC-CoMP coordinates nodes within HPN or RRH tiers, whereas inter-tier CC-CoMP coordinates RRHs with HPNs.Cross-tier coordination can burden the BBU pool, motivating partially distributed schemes.
  • Motivation and architecture: CC-CoMP uses large-scale spatial cooperative processing in the centralized BBU pool to alleviate traditional CoMP challenges.The centralized design addresses complexity, synchronization, channel estimation, and signaling burdens.
  • Challenges: Full-scale coordination requires a channel matrix covering all UEs and RRHs/HPNs, creating high computational complexity and channel-estimation overhead.The challenge arises from coordinating all network participants.
  • Challenges: Channel-matrix sparsity can support compression, but the key design question is how much compression preserves system performance.Only a small fraction of channel entries typically have reasonably strong gains.
  • Challenges: CC-CoMP scale has a trade-off: small coordination groups may yield limited SE gains, whereas large groups increase signaling, CSI, and CSI-quality burdens.CSI accuracy and instantaneity decline as coordination scale increases.
  • Design approaches: The bi-section GSBF algorithm is preferable with huge RRH counts for low complexity, while iterative GSBF suits medium-size networks for better performance.Both algorithms address joint RRH selection and power-minimization beamforming.

B. Large-Scale Cooperative Multiple Antenna Processing (LS-CMA)

LS-CMA uses large antenna arrays at HPNs to improve capacity, energy efficiency, coverage, and interference behavior, while requiring careful deployment and pilot management.

  • Architecture: LS-CMA equips HPNs with hundreds of low-power antennas and large-scale signal processing.The technique is also identified as massive MIMO.
  • Performance: Capacity increases linearly with antenna count under channel hardening, while energy-efficiency performance also improves.This follows from the law of large numbers in channel propagation.
  • Performance: 10 times or more capacity and radiated EE improvement on the order of 100 times were reported with a 100-element linear array and ideal backhaul.The comparison is against a traditional single-antenna configuration.
  • H-CRAN integration: HPNs with LS-CMA can reduce adjacent-tier interference and release constrained fronthaul by serving larger areas with fewer RRHs.Larger serving distances dilute the density of active transmitters.
  • H-CRAN integration: Inter-tier CC-CoMP can combine HPNs with LS-CMA and RRHs through cooperative beamforming to mitigate inter-tier interference and improve EE and SE.This is identified as an important H-CRAN approach.
  • Challenges: Pilot contamination affects LS-CMA strongly, although covariance-based Bayesian estimation can completely remove it under a non-overlapping dominant-covariance condition.The stated result depends on the covariance assumption.
  • Deployment trade-offs: Too many LS-CMA HPNs sacrifice RRH gains, while too few make H-CRANs resemble C-RANs, making density and site optimization necessary.The section identifies the HPN–RRH balance as a critical deployment trade-off.

C. Cloud Computing based Cooperative Radio Resource Management (CC-CRRM)

CC-CRRM applies cloud-based stochastic optimization across H-CRAN resources and timescales, while CC-SON automates network management; scalability remains a central challenge.

  • CC-CRRM: CC-CRRM jointly manages shared radio resources across RRHs and HPNs, making computational and signaling scalability a key obstacle.Its scope is broader than single-base-station radio resource management.
  • CC-CRRM: CC-CRRM adapts power, data rate, CC-CoMP, LS-CMA, scheduling, and RRH/HPN association to real-time CSI and QSI.Controls operate across PHY, MAC, and network timescales.
  • CC-CRRM: Separating timescales decomposes the stochastic control problem into lower-dimensional subproblems solvable with stochastic online learning.The decomposition is illustrated in Fig. 6.
  • CC-CRRM: Delay-aware CC-CRRM uses global QSI and CSI to capture transmission opportunities and data-flow urgency.This combines queue-state and channel-state information in resource adaptation.
  • CC-CRRM: Timescale separation reduces signaling overhead and computational complexity, while stochastic online learning provides robustness to CSI, traffic, and parameter uncertainty.The robustness claim contrasts stochastic learning with heuristic methods.
  • CC-CRRM limitations: Global state and coupled queue dynamics create a curse of dimensionality that complicates scalable CC-CRRM derivation.The challenge is especially relevant to bursty mobile data traffic.
  • CC-SON: CC-SON integrates network planning, configuration, and optimization into an automatic cloud-centered process requiring minimal manual intervention.It also targets interference mitigation and operational-cost reduction.
  • CC-SON: Hierarchical CC-SON centralizes RRH functions in the BBU pool while distributing functions between the BBU pool and HPNs.Its self-configuration, self-optimization, and self-healing functions follow this architecture.

IV. CHALLENGES AND OPEN ISSUES IN H-CRANS

H-CRAN research still faces open challenges in stochastic-geometry performance analysis, constrained resource allocation, and mobility modeling. Centralized computation and large-scale cooperation offer potential gains over HetNets and C-RANs, but tractable analytical frameworks remain incomplete.

  • Performance Analysis with Stochastic Geometry: Key unresolved challenges include theoretical analysis with stochastic geometry, optimal resource allocation under constrained fronthaul, and standard development.These challenges are identified as open issues affecting H-CRAN architecture and performance research.
  • Performance Analysis with Stochastic Geometry: Stochastic geometry is used to derive tractable closed-form expressions for coverage probability and average rate, while broader SE and EE frameworks remain needed.PPP models are tractable for key network metrics, but closed-form C-RAN SE analysis and H-CRAN EE and mobility analysis remain open.
  • Performance Analysis with Stochastic Geometry: H-CRANs make interference coordination more feasible through centralized BBU-pool computation than conventional HetNets.HetNets constrain spatial-domain interference cancellation and collaboration through cross-tier backhaul, whereas H-CRANs support larger-scale cooperative processing.
  • Performance Analysis with Stochastic Geometry: CC-CoMP and LS-CMA can provide higher performance gains through more efficient large-scale cooperative signal processing.These techniques exploit H-CRAN centralized processing capabilities beyond the cooperation feasible in HetNets.
  • Performance Analysis with Stochastic Geometry: H-CRANs can obtain additional performance gains over C-RAN through mobility enhancement, including handover success ratio and sojourn time.A tractable mobility model is needed to analytically evaluate these gains.

B. Performance Optimization of Constrained Fronthaul

Constrained fronthaul links limit H-CRAN performance through capacity shortages and latency, complicating coordinated processing and resource allocation. Advanced strategies are needed because optimizing SE or EE under these constraints is generally NP-complete.

  • Performance Optimization of Constrained Fronthaul: Non-ideal fronthaul links between RRHs and the BBU pool deteriorate overall SE and EE performance.Both limited capacity and time latency are described as inevitable practical constraints.
  • Performance Optimization of Constrained Fronthaul: Insufficient fronthaul capacity prevents RRHs from fully using available radio resources and becomes more severe with CC-CoMP, LS-CMA, and CC-CRRM.The total transmission bit rate per RRH must not exceed its corresponding fronthaul-link capacity.
  • Performance Optimization of Constrained Fronthaul: Fronthaul latency makes CSI outdated, reducing the accuracy of sparse processing and hindering optimization of SE and EE.Latency is identified as a key handicap for joint performance optimization.
  • Performance Optimization of Constrained Fronthaul: Advanced cooperative processing and resource-allocation strategies are needed to overcome constrained-fronthaul problems.The paper calls for solutions that maximize SE or EE while accounting for fronthaul constraints.
  • Performance Optimization of Constrained Fronthaul: Optimizing SE or EE under constrained fronthaul is generally an NP-complete problem.This complexity motivates research into practical optimization solutions.

C. H-CRAN Standardizations

H-CRAN standardization must extend existing HetNet and C-RAN interfaces while supporting timely, distributed coordination among HPNs, RRHs, and the BBU pool. Proposed directions include enhanced interfaces, partial local resource management, and cloud-based SON.

  • H-CRAN Standardizations: H-CRAN standardization should remain backward compatible with existing C-RAN and HetNet standards.Existing HetNet interfaces and functions provide a basis for extending H-CRAN backhaul and coordination standards.
  • H-CRAN Standardizations: RRH on/off, CC-CoMP, LS-CMA, CC-CRRM, and CC-SON are potential H-CRAN standardization issues for 3GPP Release 13 and beyond.The paper presents these functions as extensions and evolutions of HetNet capabilities.
  • H-CRAN Standardizations: Future standards should clarify HPN, RRH, BBU-pool, backhaul, and fronthaul functionalities and interfaces.These elements are emphasized to support more timely, flexible, and integrated HPN coordination with the BBU pool.
  • H-CRAN Standardizations: The X2/S1 interfaces between the BBU pool and HPNs should be enhanced beyond existing HetNet interfaces.Existing interfaces were mainly designed for adaptive interference coordination and spatial-domain joint processing in HetNets.
  • H-CRAN Standardizations: Resource allocation, RRH/HPN association, and power control should be performed partly by HPNs rather than entirely at the BBU pool.The proposed distribution aims to support H-CRAN coordination, while CC-SON should extend current HetNet SON standards.
Loading 1410.3028v1…