Source-linked AI summary
User Association in 5G Networks: A Survey and an Outlook
Dantong Liu, Lifeng Wang, Yue Chen, Maged Elkashlan, Kai-Kit Wong, Robert Schober, Lajos Hanzo
TL;DR
5G user association must support diverse technologies and objectives beyond conventional max-RSS decisions. This survey develops a taxonomy and reviews association algorithms across HetNets, massive MIMO, mmWave, and energy-harvesting networks, identifying technology-specific challenges and design opportunities.
Problem
Conventional user association is insufficient for emerging 5G networks, whose technologies create diverse challenges involving network performance, channel characteristics, energy, and traffic.
Method
The paper introduces a five-branch taxonomy and uses it to survey user association algorithms across HetNets, massive MIMO, mmWave, and energy-harvesting networks.
Results
The survey categorizes state-of-the-art algorithms and highlights how each technology’s inherent features substantially affect user association decisions.
Takeaways & Limitations
Designing sophisticated 5G user association requires accounting for technology-specific factors, including mmWave channel characteristics, massive-MIMO implementation, and energy-related tradeoffs.
Abstract
from arXiv · showhide
The fifth generation (5G) mobile networks are envisioned to support the deluge of data traffic with reduced energy consumption and improved quality of service (QoS) provision. To this end, the key enabling technologies, such as heterogeneous networks (HetNets), massive multiple-input multiple-output (MIMO) and millimeter wave (mmWave) techniques, are identified to bring 5G to fruition. Regardless of the technology adopted, a user association mechanism is needed to determine whether a user is associated with a particular base station (BS) before the data transmission commences. User association plays a pivotal role in enhancing the load balancing, the spectrum efficiency and the energy efficiency of networks. The emerging 5G networks introduce numerous challenges and opportunities for the design of sophisticated user association mechanisms. Hence, substantial research efforts are dedicated to the issues of user association in HetNets, massive MIMO networks, mmWave networks and energy harvesting networks. We introduce a taxonomy as a framework for systematically studying the existing user association algorithms. Based on the proposed taxonomy, we then proceed to present an extensive overview of the state-of-the-art in user association conceived for HetNets, massive MIMO, mmWave and energy harvesting networks. Finally, we summarize the challenges as well as opportunities of user association in 5G and provide design guidelines and potential solutions for sophisticated user association mechanisms.
I. INTRODUCTION
5G must accommodate rapidly increasing mobile traffic while improving energy consumption and QoS. This survey focuses on user association as a central mechanism across major 5G technologies and excludes re-association and handover.
- Motivation: Global mobile data traffic was projected to increase nearly tenfold from 2014 to 2019, reaching 24.3 exabytes per month.Three-fourths of the projected 2019 traffic was expected to be video.
- 5G Goals: 5G targets include higher wireless area spectral efficiency, lower energy consumption, faster service creation, dependable connectivity, and support for dense device deployments.The stated targets include 1,000 times higher area spectral efficiency and 10 times lower energy consumption per service.
- User Association: Existing max-RSS user association is inadequate for emerging 5G technologies, motivating more sophisticated algorithms.The paper relates user association to selecting a serving BS before data transmission and to network performance.
- Survey Scope: The survey reviews user association in HetNets, massive MIMO, mmWave, and energy harvesting networks using a structured taxonomy.Its contributions cover recent algorithms, technology-specific effects, challenges, and design considerations.
- Scope Boundary: The survey focuses on user association rather than the closely related re-association or handover problem.The authors explicitly leave the trigger decision for re-association or handover outside the survey’s scope.
II. TAXONOMY
The paper introduces a taxonomy for systematically examining 5G user association algorithms across five distinct design dimensions.
- Taxonomy Structure: The taxonomy organizes user association algorithms into five non-overlapping branches: Scope, Metrics, Topology, Control, and Model.Each branch is further divided into categories for evaluating algorithm design aspects.
A. Scope •
The scope dimension covers heterogeneous, massive-MIMO, mmWave, energy-harvesting, and other candidate 5G network technologies. Their physical and operational characteristics create distinct association considerations.
- HetNets: HetNets use dense low-power cells beneath macrocells to reuse spectrum, offload traffic, and improve coverage quality.Picocells, femtocells, and relays are identified as representative small-cell elements.
- Massive MIMO: Massive MIMO uses large antenna arrays to serve many single-antenna users while increasing power gain, received power, and spectrum efficiency.These properties can reduce required transmit power and improve throughput, although channel estimation errors and hardware impairments remain relevant.
- mmWave Networks: mmWave networks provide large bandwidth and directional beamforming, but high-frequency path loss restricts transmission primarily to short-range systems.Fixed-aperture arrays can pack more antennas and achieve increased array gain, with analog beamforming used under hardware constraints.
- Energy Harvesting Networks: Energy-harvesting networks can obtain power from renewable environmental sources or ambient radio signals, including through wireless power transfer.Renewable energy is intermittent, which challenges reliable QoS provision.
- Other 5G Technologies: Other candidate technologies include SON, D2D, C-RAN, and full-duplex communication, each introducing distinct association or network-management considerations.Their described capabilities include self-optimization, direct device transmission, centralized processing, and simultaneous uplink/downlink use.
B. Metrics
The metrics dimension groups user association objectives around coverage, spectrum efficiency, energy efficiency, QoS, and fairness. Studies may optimize one metric or combine several.
- Metric Categories: Five commonly used user association metrics are outage/coverage probability, spectrum efficiency, energy efficiency, QoS, and fairness.These metrics are used individually or in combination to determine BS-user assignments.
- Outage/Coverage Probability: Outage and coverage probability measure whether SINR falls below or rises above a threshold, respectively.These probabilities help benchmark the average throughput of a randomly chosen user.
- Spectrum Efficiency: Spectrum efficiency measures the maximum information rate transmitted over a given bandwidth and is required because 5G faces rising traffic and limited spectrum.
- Energy Efficiency: Energy efficiency is commonly measured as total data rate per total energy consumption or as direct power or energy savings.The ratio metric is expressed in bits/Joule.
- QoS and Fairness: QoS is quantified through measures such as traffic delay, user throughput, and SINR, while fairness concerns equitable resource allocation across users and cells.In HetNets, fairness applies both within cells and across cells in different tiers.
C. Topology
The paper distinguishes regular-grid and random-spatial topology models for 3-tier HetNets, balancing simulation realism and analytical tractability.
- Grid models place base stations on regular layouts such as hexagonal grids, but performance evaluation requires time-consuming Monte Carlo simulations and often resists mathematical analysis.
- Random spatial models capture topological randomness and enable closed-form performance expressions through stochastic geometry.
- The grid example uses macro BSs at hexagonal-cell centers, with pico and femto BSs positioned along the macro BSs.
- The random-spatial example overlays macro BSs with pico and femto BSs, represented by red circles, green triangles, and blue squares, respectively.
- Random-spatial analytical accuracy depends on whether the selected stochastic-geometry point processes represent real network conditions adequately.
D. Control
User-association control can be centralized, distributed, or hybrid, with different trade-offs among optimality, convergence, complexity, and signaling overhead.
- Centralized control uses a single entity to collect network information and assign users to BSs, enabling network-wide optimal allocation and fast convergence.
- Centralized control may impose excessive signaling overhead in medium- to large-sized networks.
- Distributed control lets BSs and users make autonomous association decisions, offering low implementation complexity and signaling overhead for large HetNets.
- Hybrid control combines centralized and distributed mechanisms, such as distributed BS power control with centralized network-wide load balancing.
E. Model
The survey organizes user-association models around utility, game theory, combinatorial optimization, and stochastic geometry, while highlighting assumptions and tractability limits.
- Utility represents the satisfaction a service provides to a decision maker and may encode spectrum efficiency, energy efficiency, or QoS.
- Game-theoretic models support distributed, low-overhead self-configuration but assume rational BSs and users pursuing their own interests.
- Combinatorial optimization models binary user–BS associations and resource constraints, making exhaustive optimal search NP-hard and computationally prohibitive for medium-sized networks.
- Relaxing binary associations to continuous values can enable convex and dual analysis, but may create a duality gap.
- Stochastic geometry captures random network geometry and yields tractable analytical metrics, yet selecting point processes that accurately represent 5G networks remains an open challenge.
- In dense HetNets, max-RSS can associate most users with macrocells because of transmit-power disparities, motivating biased association to offload users toward small cells.
A. User Association for Outage/Coverage Probability Optimization
The survey reviews outage and coverage optimization alongside spectrum- and energy-efficiency objectives, emphasizing trade-offs among coverage, fairness, load, and power consumption.
- Outage/Coverage Probability: Outage and coverage probability are used to evaluate desired-user performance and are primary metrics for stochastic-geometry analyses of user association.
- Outage/Coverage Probability: Adding lightly loaded femtocells and picocells can increase overall coverage probability, while heterogeneous loads require activity-factor-aware interference modeling.
- Outage/Coverage Probability: Stochastic-geometry studies determine biases producing the highest SIR or rate coverage, and also examine biasing with macrocell–small-cell spectrum partitioning.
- Spectrum Efficiency: Maximizing sum data rate can produce unfair allocation: overloaded small cells leave noncentral users starved while privileged macrocell-center users obtain high rates.
- Spectrum Efficiency: Distributed fractional association can transform an intractable primal problem into a convex optimization and converge toward an optimum under a bounded discrepancy.
- Cross-metric objectives: Joint user association with channel allocation, transmission coordination, or BS sleep modes is studied to optimize network performance and energy efficiency.
- Energy Efficiency: Energy-efficiency formulations minimize consumption under traffic demands or maximize total data rate divided by total network energy consumption.
D. User Association Accommodating Other Emerging Issues in HetNets
HetNets create user-association challenges beyond transmit-power disparity, particularly uplink-downlink asymmetry, backhaul limits, and mobility. Existing work has begun addressing these issues, but backhaul-aware mechanisms remain important.
- Uplink-Downlink Asymmetry: Uplink-downlink asymmetry makes single-direction association potentially ineffective for the opposite link.HetNets differ across channel quality, traffic, coverage, and hardware constraints; coverage asymmetry is especially severe.
- Uplink-Downlink Asymmetry: Joint uplink and downlink optimization can maximize admitted users while minimizing weighted uplink power, although performance depends on heuristic weighting.The cited objective was later improved to maximize a different utility under the same coupled design.
- Backhaul Bottleneck: Most studies assume perfect backhaul, overlooking how backhaul implementation and capacity constrain wireless-front-end gains.This motivates association mechanisms that explicitly incorporate backhaul capacity.
- Backhaul Bottleneck: Backhaul-aware approaches optimize spectrum efficiency, user-rate utility, weighted sum rate, or network capacity under capacity and load constraints.Examples include distributed relaxed combinatorial optimization, waterfilling-like association, and heuristic capacity maximization.
- Mobility Support: Small-cell densification reduces footprints and can increase handovers for moderately or highly mobile users unless mobility is incorporated.This creates a need for association decisions that improve long-term system performance while avoiding excessive handovers.
3) Mobility Support:
Mobility, multiple performance objectives, and practical deployment gaps complicate user association in emerging networks. Massive MIMO expands coverage and association likelihood for macrocells, while energy-efficient and theoretically grounded designs remain active research areas.
- Mobility Support: Speed-dependent biasing can improve coverage probability and overall network performance under mobility.Stochastic-geometry analysis relates both the optimal bias and coverage probability to user speed.
- Mobility Support: Dual connectivity improves mobility resilience and attainable throughput by allowing simultaneous macrocell and small-cell association.Its association criteria and sum-rate optimization have been studied through simulations and low-complexity algorithms.
- Summary and Discussions: QoS, spectrum efficiency, energy efficiency, and fairness are essential for real-time applications, but most existing work does not address all four together.Closed-form analysis of their tradeoffs remains difficult in complex HetNets.
- Summary and Discussions: Uplink-downlink asymmetry, backhaul bottlenecks, and mobility support remain immature topics requiring further theoretical analysis and independent practical verification.Most prior contributions focus on optimization or theoretical performance analysis rather than realistic systems.
- Received Power Based User Association: Massive MIMO increases macrocell association probability because large array gains persist even when macro and pico transmit powers are equal.Increasing macrocell antenna count further raises association probability and may reduce the number of required small cells.
B. User Association for Spectrum Efficiency Optimization
Massive MIMO changes spectrum- and energy-efficiency considerations in user association through array gains, multiplexing, interference suppression, and circuit-power costs. The literature indicates improved per-user energy efficiency and fairness, while association must balance antenna count against served users.
- Spectrum Efficiency Optimization: Max-RSS association may fail to balance load when multi-tier base stations use different antenna counts and zero-forcing beamforming.Massive MIMO can therefore make received-power association inadequate for load balancing.
- Energy Efficiency Optimization: Circuit-power models show that per-user energy efficiency can improve despite hundreds of antennas when regularized zero-forcing serves more users while suppressing interference.Practical models include both RF and circuit-power consumption.
- Energy Efficiency Optimization: Lower macrocell transmit power yields higher energy efficiency for a fixed antenna count, indicating that massive MIMO can meet QoS at reduced transmit power.The cited trend is shown in energy-efficiency-versus-antenna-count analysis.
- Energy Efficiency Optimization: Geometric-mean energy-efficiency maximization improves low-efficiency users and produces more uniform energy efficiency through proportional fairness.The proposed algorithm’s CDF approaches max-SINR association at 4 × 10^5 bits/Joule.
- Energy Efficiency Optimization: Energy-efficient association in massive-MIMO HetNets remains immature and must balance the number of antennas against the number of served users.Further research is needed to translate these results into practical 5G networks.
V. USER ASSOCIATION IN MMWAVE NETWORKS
MmWave channel loss, blockage, propagation differences, mobility, and array gains make user association highly dependent on link conditions rather than distance or RSS alone. The survey identifies broad open challenges spanning bandwidth, energy, mobility, and coupled access.
- MmWave Channel Characteristics: MmWave links suffer high path loss, stronger NLOS attenuation, and greater blockage sensitivity, making user association central to coverage and throughput.Indoor users may not be adequately covered by outdoor mmWave base stations.
- MmWave Channel Characteristics: A farther base station may provide a better directional link than the geographically closest base station.This follows from directional propagation and heterogeneous channel conditions.
- User Association Challenges: RSS-based association can waste resources and cause frequent handovers with increased reassociation overhead and delay.These limitations are identified in current mmWave standards and related studies.
- MmWave HetNet Integration: MmWave cells can offload existing HetNets and reduce interference because they operate at higher frequencies than conventional cellular tiers.The mmWave tier is treated as an additional tier in future 5G HetNets.
- Open Challenges: Future association methods should incorporate LOS/NLOS propagation, blockage, mobility, bandwidth, array gains, and energy consumption.Large bandwidth and antenna arrays improve throughput but also require careful power consideration.
- Coupled and Decoupled Access: Downlink-uplink decoupling is difficult in mmWave networks because beamforming tends to rely on channel reciprocity.One proposed arrangement associates downlink access with an mmWave base station and uplink access with a sub-6 GHz macrocell.
- Open Challenges: Efficient user association for 5G mmWave networks remains a promising research field with numerous open facets.The survey frames mmWave association as an enabling component requiring fundamental research.
VI. USER ASSOCIATION IN ENERGY HARVESTING NETWORKS
Energy harvesting networks require user association to account for renewable-energy availability, load variation, and energy cooperation. The survey organizes existing work while identifying tradeoffs and open challenges for these settings.
- Summary of Existing Work: Existing energy-harvesting research is summarized qualitatively across renewable-energy-powered network scenarios.The survey provides comparison and challenge summaries in Tables VI and VII.
- Renewable Energy Powered Networks: Renewable-energy-powered base stations introduce user-association challenges because harvested energy varies across time and space.Association decisions must adapt to both energy and load variations.
- Renewable Energy Powered Networks: Hybrid energy sources combine grid power with renewable sources to support uninterrupted service despite intermittent harvesting.Hybrid-powered base stations are presented as preferable to systems relying solely on renewable energy.
- Energy Cooperation Enabled Networks: Energy cooperation lets base stations with excess harvested energy assist stations facing deficits through renewable-energy transfer.This can improve renewable-energy utilization and reduce grid-energy consumption.
- Energy Cooperation Enabled Networks: User association in energy-cooperation networks trades traffic offloading against energy transfer and remains an open research problem.Associating users with a farther, energy-rich base station can avoid transfer losses but may reduce received signal strength.
B. User Association in RF WPT Enabled Networks
RF wireless power transfer offers environment-independent and schedulable energy harvesting for cellular devices. The survey distinguishes downlink and uplink association objectives while noting important research gaps and broader integration opportunities.
- RF WPT Motivation: RF WPT can prolong device lifetimes while operating independently of environmental conditions and allowing flexible scheduling.Its use also raises concerns about potentially harmful interference.
- Ambient RF Energy Harvesting: Ambient RF harvesting supports uplink transmission through a harvest-then-transmit strategy in K-tier cellular networks.The surveyed approach associates users based on the best available criterion described in the cited work.
- RF WPT Constraints: Ambient RF scavenging is sufficient only for small sensors, so larger devices require dedicated power beacons.This limitation establishes a practical boundary for relying on ambient RF energy alone.
- Association Objectives: WPT association designs target either maximum harvested downlink energy or minimum uplink path-loss.Uplink-downlink decoupling is identified as a promising way to pursue both objectives.
- Open Directions: User association combining renewable-powered base stations with RF-harvesting users remains unexplored and is identified as promising for 5G.The survey also describes energy-cooperation-enabled networks as an open field.
B. Device-to-Device Communication
D2D communication enables direct device transmission while retaining cellular-network control, but its interaction with other 5G technologies complicates user association. The survey identifies limited research and emphasizes scenario-specific metric choices.
- Device-to-Device Communication: D2D communication enables proximity-based direct transmission without assistance from the base station or core network.The cellular network remains involved in the control plane, and operation may be inband or outband.
- Related 5G Technologies: C-RAN centralizes and coordinates baseband processing among sites, reducing CAPEX and OPEX while mitigating inter-RRH interference.These architectural features create additional considerations for joint user association.
- Related 5G Technologies: Full-duplex communication can potentially double spectrum efficiency, but self-interference suppression is critical to reception quality.User-association analysis must account for the operating mode and interference-management conditions.
- Research Gaps: User association mechanisms for SONs, D2D, C-RAN, and full-duplex networks remain limited and require further research.These technologies impose distinct design requirements despite potential efficiency and cost benefits.
- Design Considerations: The survey uses outage/coverage probability, bandwidth efficiency, energy efficiency, QoS, and fairness as user-association metrics.Metric selection should reflect the requirements of the specific service and 5G scenario.