PaperlayerView exact revision on arXiv

Source-linked AI summary

A Survey on Device-to-Device Communication in Cellular Networks

Arash Asadi, Qing Wang, Vincenzo Mancuso

arXiv:1310.0720v6cs.GTcs.ITcs.NI

TL;DR

D2D communication addresses rising cellular data demand by enabling direct communication between nearby users, but interference control and protocol design remain open problems. This paper surveys the literature using a taxonomy based on communicating spectrum, reviews existing approaches, and identifies immature areas and future research directions. Its conclusion is that underlay, overlay, and outband designs involve distinct interference, resource-allocation, and QoS trade-offs, while the field remains immature with numerous open issues.

  • Problem

    Rising mobile data demand motivates D2D communication, but interference control, protocol design, and the lack of a survey remain research gaps.

  • Method

    The paper extensively surveys D2D literature and categorizes it by spectrum into inband underlay, inband overlay, and outband controlled or autonomous communication.

  • Results

    The survey identifies power control and interference management as central underlay issues, dedicated-resource efficiency as an overlay concern, and uncontrollable unlicensed-spectrum interference as an outband QoS challenge.

  • Takeaways & Limitations

    D2D communication in cellular networks remains immature, with interference management, power control, and other open issues requiring further research.

Abstract

from arXiv · show

Device-to-Device (D2D) communication was initially proposed in cellular networks as a new paradigm to enhance network performance. The emergence of new applications such as content distribution and location-aware advertisement introduced new use-cases for D2D communications in cellular networks. The initial studies showed that D2D communication has advantages such as increased spectral efficiency and reduced communication delay. However, this communication mode introduces complications in terms of interference control overhead and protocols that are still open research problems. The feasibility of D2D communications in LTE-A is being studied by academia, industry, and the standardization bodies. To date, there are more than 100 papers available on D2D communications in cellular networks and, there is no survey on this field. In this article, we provide a taxonomy based on the D2D communicating spectrum and review the available literature extensively under the proposed taxonomy. Moreover, we provide new insights to the over-explored and under-explored areas which lead us to identify open research problems of D2D communication in cellular networks.

I. INTRODUCTION

D2D communication lets nearby mobile users communicate directly, offering potential network-performance benefits as data-intensive applications increase demand. The field has expanded across use-cases and spectrum arrangements, but interference, protocol, and standardization challenges remain active research concerns.

  • Motivation: Growing demand from data-intensive services motivates D2D as a potential component of next-generation cellular technologies.Examples include video sharing, gaming, and proximity-aware social networking.
  • D2D communication: D2D communication directly connects two mobile users without traversing the base station or core network.It may use either cellular spectrum (inband) or unlicensed spectrum (outband).
  • Potential benefits: D2D can potentially improve spectral efficiency, throughput, energy efficiency, delay, and fairness.Direct communication is especially relevant when users are close enough to communicate without the base station.
  • Use-cases and development: Research has introduced use-cases including multihop relays, multicasting, peer-to-peer communication, video dissemination, machine-to-machine communication, and cellular offloading.FlashLinQ and 3GPP ProSe represent implementation and standardization efforts for D2D-enabled cellular networks.
  • Spectrum arrangements: Inband D2D is categorized as underlay, where cellular and D2D links share resources, or overlay, where D2D receives dedicated cellular resources.Underlay emphasizes interference mitigation, while overlay makes resource allocation important because dedicated resources may be wasted.
  • Related architectures: D2D differs from MANET and cognitive radio networks because cellular infrastructure is expected to oversee its control plane, often through an eNB.D2D is mainly intended for single-hop proximity connectivity rather than MANET-style multihop routing.

B. Contributions and Organization of the Survey

The survey introduces a spectrum-based taxonomy for D2D communications, reviews inband and outband literature, and identifies performance trade-offs and open research issues. It organizes findings around interference management, resource allocation, mode selection, and network coding, while noting that higher gains can require greater computational overhead.

  • Taxonomy: The survey categorizes D2D literature by communicating spectrum and reviews inband and outband approaches under this taxonomy.Inband communication is further divided into underlay and overlay categories.
  • Taxonomy: Underlay inband D2D reuses cellular resources but creates cellular–D2D interference that requires complex resource-allocation methods.Overlay assigns dedicated cellular resources to D2D links, while outband D2D uses unlicensed spectrum and an extra interface.
  • Surveyed approaches: The surveyed literature uses interference management, mode selection, resource allocation, and network coding to improve cellular-network performance.The reviewed studies address spectrum efficiency, energy efficiency, coverage, capacity, and other targets.
  • Reported findings: D2D interference-control proposals report gains including 129% over conventional management, up to 374% higher average throughput without cancellation, and up to 650% higher cell throughput in favorable placement.Other studies report up to 65% higher throughput, a 50% system-throughput gain, and capacity-region enlargement of up to 60%.
  • Reported findings: Network coding with complementary-user grouping increases capacity by 34% versus random selection and 16% versus decode-and-forward relaying.The same study reports that multi-antenna capability increases the number of D2D users by 30%.
  • Trade-offs and open issues: Advanced mathematical techniques can produce higher gains but also increase computational overhead, whereas self-organized methods introduce less overhead than network-controlled methods.The survey identifies open issues and research directions alongside discussion of assumptions, maturity, and real-world emergence.

B. Power Efficiency

The surveyed work improves power efficiency mainly through dynamic mode selection and power allocation, balancing performance gains against computational cost.

  • Around 20% lower downlink power consumption was achieved than traditional OFDMA without D2D by integrating heuristic power allocation and mode selection.The heuristic first allocates cellular resources, then selects resources and modes for D2D users subject to a power threshold.
  • Up to 100% gain over a pathloss-based mode-selection scheme was reported for an algorithm choosing the mode with higher measured power efficiency.Power efficiency is computed from transmission rate and power consumption, but exhaustive search over device-mode combinations is required.
  • Joint mode selection, resource allocation, and power allocation can be formulated as a strongly NP-hard linear-programming problem.The cited work simplifies the problem by considering power allocation in a single-cell setting.
  • The surveyed power-efficiency methods dynamically switch between cellular and D2D modes, using heuristics or brute force with different computational requirements.The brute-force method performs better than the heuristic methods but requires substantially more computation.

C. Performance with QoS/Power Constraints

Research on QoS- and power-constrained D2D systems uses optimization and heuristic scheduling methods to improve performance while satisfying user requirements. These gains often come with computational or scalability limitations.

  • Up to 70% throughput gain over three compared algorithms was reported for resource allocation combining admission control, D2D power optimization, and weighted matching.The method guarantees QoS requirements for both D2D and cellular users.
  • A particle-swarm method for throughput maximization with minimum data-rate requirements achieved a 15% throughput gain over orthogonal resource sharing.
  • Scheduling and mode-selection methods also optimize mean sum-rate under QoS satisfaction using stochastic sub-gradient algorithms and CSI-aware opportunistic scheduling.
  • A two-phase low-complexity solution replaces an NP-hard integer-programming formulation for cellular and D2D resource allocation.It allocates cellular resources first, then heuristically assigns D2D subchannels while accounting for D2D power budgets.
  • QoS- and power-constrained performance improvement commonly relies on stochastic optimization, nonlinear programming, or integer optimization.These methods can improve system performance but may be unsuitable for time-stringent applications with limited computational capacity.
  • Closed-form optimal solutions reduce computational complexity, but the corresponding evaluation scenario contains only one cellular user and one D2D pair.

D. Miscellaneous

The surveyed miscellaneous work applies D2D communication to spectrum utility, coverage, fairness, reliability, M2M traffic, multicast feedback, and content distribution. These studies use mode selection, optimization, relaying, and network assistance for diverse cellular objectives.

  • Up to 33% and 500% spectrum-utility improvements were reported versus solely BS mode and solely D2D mode, respectively.The method independently selects modes and uses BS-broadcast decisions to inform future selections.
  • D2D can enlarge cellular coverage by assigning nearby nodes as virtual infrastructure or relay nodes according to network conditions and traffic requirements.
  • Reliability methods dynamically choose among receive modes using outage-probability calculations, while distributed power control limits interference through predefined cellular-user margins.
  • A Hungarian-algorithm-based heuristic can perform as well as the optimal solution while maximizing admitted D2D links and minimizing average D2D interference.
  • D2D supports M2M communication to address massive device populations and the congestion and contention limits of random medium access.
  • Clustered HARQ reduces device feedback by routing ACK/NACK status through cluster headers and using D2D retransmissions for users that miss multicast packets.The method improves multicast performance by reducing the frame-loss ratio of feedback.
  • A location-aware content-distribution scheme directs a nearby user with cached content to transmit it over D2D, avoiding retransmission by the BS.
  • The surveyed use-cases include proximity peer-to-peer gaming and social networking, alongside spectrum utilization and fairness objectives.

IV. OVERLAYING INBAND D2D

Overlay inband D2D dedicates cellular resources to D2D transmissions, reducing interference concerns but leaving fewer resources for cellular communication. The surveyed work emphasizes scheduling, power control, and relaying applications.

  • Overlaying inband D2D: Dedicated D2D resources eliminate interference from D2D transmissions to cellular transmissions but reduce resources available for cellular communication.
  • Overlaying inband D2D: BS-assisted scheduling and D2D power control are proposed to reduce D2D interference when resources are dedicated to D2D.
  • Overlaying inband D2D: Incremental relay transmission has the BS receive a D2D multicast copy and retransmit it to the receiver when the D2D transmission fails.The cited work claims this improves throughput and reduces outage probability.
  • Overlaying inband D2D: D2D retransmissions within multicast clusters use successfully decoding members to deliver packets to members that failed to decode them.
  • Overlaying inband D2D: The overlay literature is summarized in tables organized around D2D use cases and resource direction, including uplink.
  • Overlaying inband D2D: System throughput and energy efficiency are identified as performance objectives for the surveyed overlay D2D literature.
  • Overlaying inband D2D: 90% less spectrum resources were consumed than in the scenario with only one retransmitter by the proposed multicast algorithm.
  • Overlaying inband D2D: Relaying methods use either the BS as a backup retransmitter or multiple D2D users as multicast retransmitters, with low complexity reported for both.

V. OUTBAND D2D

Outband D2D uses a frequency band separate from the cellular spectrum, avoiding interference between D2D and cellular communications. It may be controlled by the cellular network or operate autonomously.

  • Outband D2D operates on a frequency band that does not overlap with the cellular spectrum.
  • Because the spectra do not overlap, outband D2D avoids interference between D2D and cellular communications.
  • Outband D2D can be managed by the cellular network or operate independently in autonomous mode.

A. Controlled

Controlled outband D2D uses cellular-network management to improve communication efficiency and reliability. Research addresses channel contention, clustering, protocol integration, and cached-content distribution.

  • Controlled: Controlled outband D2D uses cellular-network management features to improve throughput, power efficiency, multicast, and related performance measures.
  • Controlled: Grouping D2D users by QoS and restricting channel contention to one user per group is proposed to reduce simultaneous D2D and WLAN contention.
  • Controlled: Clustering cellular users lets the highest-cellular-quality cluster head communicate with the BS and forward traffic to other members.
  • Controlled: A protocol stack connecting LTE and WiFi Direct addresses messaging, signaling, channel feedback, scheduling, and security.
  • Controlled: Up to 50% lower packet delay was reported for a Round Robin scheduler, with delays below 10 ms at 99% probability.
  • Controlled: Caching popular videos on smartphones and transmitting locally available files over an unlicensed D2D band targets improved video throughput.
  • Controlled: Asynchronous content reuse combines D2D communication and mobile-device caching to maximize per-user throughput under an outage-probability constraint.

B. Autonomous

Autonomous D2D reduces cellular-network overhead without requiring BS changes, while protocol research addresses traffic spreading, LTE integration, and device discovery. Reviewed proposals target delay, energy, offloading, and direct connectivity.

  • Autonomous: Autonomous D2D requires no BS changes and can be deployed easily, but relatively few works address this category.
  • Autonomous: A BS-transparent dispatching policy spreads traffic requests among users to balance BS backlogs under dynamic traffic loads.
  • Autonomous: The proposed file-dispatching methodology reduces file-transfer delay by up to 50% and uses 80% less power than performance-centric algorithms.
  • Autonomous: The outband D2D literature includes work on interference, resource allocation, power allocation, and protocol stacks for inband and outband communication.
  • Autonomous: An inband D2D architecture adds a D2D server that manages identifiers, call establishment, capabilities, services, and mobility.
  • Autonomous: An outband protocol connects LTE and WiFi Direct by encapsulating LTE PDCP PDUs into WiFi packets for D2D transmission and relaying.
  • Autonomous: The outband protocol addresses registration, connection establishment, bearer setup, mobility, CSI reporting, scheduling, and security while minimizing changes to existing protocols.
  • Autonomous: 3GPP ProSe supports direct-mode data exchange between UEs and locally routed data paths through eNBs, alongside use cases including discovery and traffic offloading.

VII. DISCUSSIONS AND FUTURE WORK

The reviewed D2D literature commonly assumes channel-state awareness and negligible inter-cluster interference, while often using throughput under continuously backlogged traffic. These assumptions may be problematic in dense or dynamic settings.

  • Many studies assume the BS knows instantaneous CSI for cellular and D2D links so it can schedule cellular and D2D users.
  • Clustering studies commonly assume different clusters are far enough apart to have no or negligible mutual interference.
  • The no-interference assumption may not hold in populated areas or dense deployments.
  • Most papers assume the BS or D2D users always have traffic to send and therefore use throughput as a common metric.

B. Inband or Outband?

D2D research primarily considers reusing cellular spectrum or using unlicensed spectrum, each offering distinct interference, control, and allocation trade-offs. The literature remains relatively immature, with limited analytical and realistic evaluation methods.

  • Spectrum choices: Inband D2D reuses cellular resources, while outband D2D uses unlicensed spectrum; outband research has recently attracted increasing attention.Inband studies dominate the literature, but outband D2D is considered a viable alternative as devices increasingly support multiple wireless interfaces.
  • Inband D2D: Underlay inband D2D improves spectral efficiency through spatial reuse but requires interference management between D2D and cellular users.Overlay avoids overlap by dedicating cellular resources to D2D, making resource allocation important and potentially reducing spectrum efficiency.
  • Outband D2D: Outband D2D avoids cellular–D2D interference and simplifies resource allocation, but unlicensed-spectrum congestion and coordination remain disadvantages.Its feasibility also depends on devices supporting multiple wireless interfaces and on handover across different platforms.
  • Research maturity: The field is relatively young: most studies propose architectures or heuristics, while some optimization problems remain unsolved because of NP-hardness.The authors identify room for optimal solutions for interference coordination, power management, and mode selection.
  • Research maturity: Most evaluations use numerical methods or simple home-grown simulators; no study uses experimental evaluation, and popular network simulators rarely support D2D.The authors attribute the lack of testbeds partly to their high cost and limited D2D support.

E. D2D Implementation Challenges in Real World

D2D adoption faces implementation challenges spanning interference, power and resource control, radio compatibility, channel measurement, energy consumption, and feedback design. These challenges differ between inband and outband operation but collectively constrain reliable real-world deployment.

  • Core challenges: Interference management must prevent inband D2D reuse of uplink or downlink resources from disrupting cellular services.The characteristics of D2D interference are not yet well understood, so power and resource allocation remain central mechanisms.
  • Core challenges: Power allocation must protect cellular links while maintaining the D2D receiver’s minimum SINR requirement in inband operation.In outband operation, cellular–D2D interference is absent, but increasing ISM-band occupancy makes power control important for limiting congestion and collisions.
  • Core challenges: Resource allocation should assign inband resource blocks away from nearby interfering cellular UEs and avoid occupied ISM bands for outband D2D.The relevant interference sources differ across licensed and unlicensed spectrum.
  • Core challenges: D2D radio compatibility requires OFDMA transmitters for downlink-resource use and SC-FDMA receivers for uplink-resource use.Existing LTE UEs conventionally use OFDMA reception downlink and SC-FDMA transmission uplink.
  • Core challenges: Accurate channel information is required for interference management, power allocation, and resource allocation involving D2D and cellular links.D2D requires channel gains beyond the conventional information collected in cellular systems.
  • Core challenges: Device discovery creates an energy–latency trade-off: frequent listening or discovery transmissions can significantly reduce UE battery life.The protocol must balance discovery speed against UE power consumption.
  • Reliability: HARQ can improve robustness through direct receiver-to-transmitter feedback or indirect feedback routed through the eNB.Direct HARQ creates less eNB overhead and delivers ACK/NACK messages with shorter delay.

F. Potential Future Work

Future D2D research should strengthen theory, architecture, applications, and performance evaluation. The survey’s taxonomy and findings indicate that unresolved interference, power, spectrum, and realism issues remain central to the field’s development.

  • Theoretical work: Theoretical work should develop optimal mode selection, interference and power control, queue-stability analysis, utility functions, and delay bounds.The survey highlights stochastic Lyapunov optimization as one possible analytical direction.
  • Architecture: D2D architecture needs further study of centralized support for discovery, connection setup, registration, interference control, resource allocation, and security.The authors also identify software-defined networking and heterogeneous networks as relevant architectural contexts.
  • Application: Additional applications should be explored beyond relaying, multicasting, peer-to-peer communication, video dissemination, M2M communication, and cellular offloading.The survey presents application expansion as an open direction for telecommunications research.
  • Performance analysis: Realistic evaluation should use established network simulators or experimental testbeds because numerical and home-grown simulations rely on simplified assumptions.The authors specifically mention NS3, OPNET, and Omnet++ as possible evaluation platforms.
  • Survey synthesis: The survey organizes D2D literature into inband and outband categories, with underlay and overlay subcategories for inband and controlled and autonomous subcategories for outband.This taxonomy structures the review according to the spectrum used for D2D transmission.
  • Survey synthesis: Underlay primarily faces D2D–cellular power and interference control, whereas overlay avoids overlap but uses dedicated cellular resources and has lower spectral efficiency than underlay.Outband avoids D2D–cellular interference and power-control issues, but uncontrollable unlicensed-spectrum interference complicates QoS in saturated areas.
  • Conclusion: The literature review identifies D2D communication in cellular networks as immature, with numerous open issues and research directions for understanding real-world potential.The authors characterize their work as an extensive survey that exposes weaknesses, strengths, and under-explored problems.
Loading 1310.0720v6…