PaperlayerView exact revision on arXiv

Source-linked AI summary

UAV-Aided Offloading for Cellular Hotspot

Jiangbin Lyu, Yong Zeng, Rui Zhang

arXiv:1705.09024v4cs.IT

TL;DR

Cell-edge mobile terminals become a performance bottleneck when distant from a heavily loaded ground base station. The paper optimizes UAV offloading and spectrum sharing, finding that the hybrid design improves throughput over GBS-only and small-cell schemes.

  • Problem

    Cell-edge mobile terminals suffer poor channels and create a performance bottleneck, particularly during hotspot periods.

  • Method

    The paper jointly optimizes UAV trajectory, spectrum allocation, and user partitioning for hybrid cellular offloading with interference suppression.

  • Results

    The proposed hybrid network significantly improves throughput over conventional GBS-only and multiple-microcell or small-cell offloading schemes.

  • Takeaways & Limitations

    UAV offloading provides a throughput-enhancing alternative for serving cell-edge users within the proposed hybrid network design.

Abstract

from arXiv · show

In conventional terrestrial cellular networks, mobile terminals (MTs) at the cell edge often pose a performance bottleneck due to their long distances from the serving ground base station (GBS), especially in hotspot period when the GBS is heavily loaded. This paper proposes a new hybrid network architecture by leveraging the use of unmanned aerial vehicle (UAV) as an aerial mobile base station, which flies cyclically along the cell edge to offload data traffic for cell-edge MTs. We aim to maximize the minimum throughput of all MTs by jointly optimizing the UAV's trajectory, bandwidth allocation and user partitioning. We first consider orthogonal spectrum sharing between the UAV and GBS, and then extend to spectrum reuse where the total bandwidth is shared by both the GBS and UAV with their mutual interference effectively avoided. Numerical results show that the proposed hybrid network with optimized spectrum sharing and cyclical multiple access design significantly improves the spatial throughput over the conventional GBS-only network; while the spectrum reuse scheme provides further throughput gains at the cost of slightly higher complexity for interference control. Moreover, compared to the conventional small-cell offloading scheme, the proposed UAV offloading scheme is shown to outperform in terms of throughput, besides saving the infrastructure cost.

I. INTRODUCTION

The paper proposes a UAV-aided hybrid cellular architecture to offload heavily loaded GBSs by serving cell-edge MTs, whose poor channel conditions create a hotspot bottleneck. It jointly optimizes UAV trajectory, spectrum sharing, multiple access, and user partitioning to maximize common throughput, with numerical results showing gains over GBS-only and micro-cell alternatives.

  • Cell-edge MTs suffer poor channels from long GBS distances, becoming a performance bottleneck during hotspots when the GBS is heavily loaded.They may require more bandwidth or higher transmit power to match other MTs.
  • The proposed hybrid architecture adds a UAV aerial mobile BS that cyclically follows the cell edge to serve cell-edge MTs and offload the GBS.MTs are partitioned between the UAV and GBS; the UAV uses a fixed-altitude circular trajectory and cyclical time division.
  • The design maximizes the minimum common throughput of all MTs by jointly optimizing bandwidth allocation, user partitioning, and UAV circular trajectory radius.The resulting joint optimization problem is non-convex and is addressed by sequential trajectory-radius and allocation/partitioning optimization.
  • Orthogonal spectrum sharing partitions total bandwidth between the GBS and UAV, while spectrum reuse shares the whole pool for concurrent communications.Reuse requires interference suppression using directional UAV antennas and adaptive directional GBS transmission.
  • Spectrum reuse further improves spectrum efficiency and common throughput over orthogonal sharing, but increases practical interference-avoidance complexity.The tradeoff arises from mutual interference between concurrent GBS and UAV communications.
  • The optimized UAV hybrid network greatly improves spatial throughput over GBS-only networks and supports higher user density under identical target-rate requirements.A single UAV/mobile BS significantly outperforms micro-cell offloading in throughput while saving infrastructure cost.

II. SYSTEM MODEL

The system model combines a GBS and UAV to serve ground MTs, partitioning users by distance and using cyclical UAV access for cell-edge MTs. It models fixed circular UAV flight, directional UAV coverage, distinct GBS/UAV channels, and both orthogonal sharing and spectrum reuse.

  • User association and access: MTs are partitioned by threshold rI into inner-region KG users served by the GBS and exterior-ring KU users served by the UAV.The partition is based on distance to the GBS, with KG covering the inner disk of radius rI and KU the remaining exterior ring.
  • UAV mobility: The UAV flies at fixed altitude HU and constant speed V along a circular trajectory centered over the GBS, with period T = 2πrU/V.The circular path periodically serves cell-edge users and is considered energy-efficient for UAV flight.
  • User association and access: A time-division cyclical multiple access scheme schedules cell-edge MTs near the UAV’s current position to exploit their strongest communication links.Different cell-edge MTs communicate with the UAV cyclically as it flies close to them.
  • Channel and coverage model: The UAV uses a directional antenna with ground main-lobe coverage radius rc = HU tan ΦU, while scheduled MTs are assumed to remain within coverage.Increasing beamwidth ΦU reduces main-lobe antenna gain, so ΦU and scheduled users must be carefully designed.
  • Spectrum sharing: The model considers orthogonal spectrum sharing without interference and spectrum reuse with concurrent transmissions whose mutual interference is effectively suppressed.Spectrum reuse improves spectrum efficiency but is more complicated to design and implement in practice.

III. ORTHOGONAL SPECTRUM SHARING · A. UAV-MT Communication · 1) Average throughput:

The orthogonal spectrum-sharing analysis derives UAV-MT and GBS-MT throughputs and maximizes the common normalized minimum throughput by jointly optimizing UAV trajectory, user partitioning, and bandwidth allocation. For each cell-edge MT, average throughput over a UAV flying period depends on association duration and instantaneous UAV communication rate.

  • III. ORTHOGONAL SPECTRUM SHARING: The orthogonal spectrum-sharing scheme derives the achievable throughputs of UAV-MT and GBS-MT communications.The common minimum throughput is normalized by the total system bandwidth W.
  • III. ORTHOGONAL SPECTRUM SHARING: The optimization maximizes the common minimum throughput ¯ν by jointly selecting UAV trajectory radius rU, user partitioning radius threshold rI, and bandwidth allocation.The objective is defined over all mobile terminals and uses normalized throughput in bits per second per Hz (bps/Hz).
  • A. UAV-MT Communication: For each MT k, association time τk is the total duration during which it communicates with the UAV within each flying period T.The association interval is bounded by starting time ts,k and ending time te,k, with τk = te,k − ts,k.
  • 1) Average throughput:: The average throughput of cell-edge MT k ∈ KU over period T is determined by τk and its instantaneous UAV communication rate.The average is formed over the interval in which MT k is associated with the UAV.
  • 1) Average throughput:: The UAV allocates transmit power pk(t) to MT k ∈ KU(t), whose instantaneous achievable rate Rk(t) is expressed in bps/Hz.The rate is specified during MT k’s association time.
  • 1) Average throughput:: Rk(t) depends on allocated transmit power pk(t), UAV-MT horizontal link distance dk(t), and normalized per-user bandwidth bU(t).The bandwidth bU(t) depends on the number of MTs |KU(t)| associated with the UAV at time t.
  • 1) Average throughput:: The UAV-MT association pattern specifies KU(t) over 0 ≤ t ≤ T together with the associated distance dk(t) and transmit power pk(t).These quantities are designed over each MT’s association interval ts,k ≤ t ≤ te,k.

2) Power allocation: · 3) UAV-MT association: · 4) Lower bound of average throughput:

The paper fixes equal UAV transmit-power allocation among associated MTs and uses a cyclical ring-segment association rule for cell-edge users. It then derives a common lower bound on UAV-served average throughput, showing that smaller association angles improve the bound but may leave the UAV unassociated at times.

  • 2) Power allocation:: The UAV allocates equal transmit power to its associated MTs at each time instant, with each user receiving p_k(t) = P_U/|K_U(t)|.P_U denotes the UAV’s maximum transmit power.
  • 2) Power allocation:: The association K_U(t) determines average throughput through normalized per-user bandwidth and each MT’s association time period.The per-user bandwidth is b_U(t) = ρ/|K_U(t)|.
  • 3) UAV-MT association:: MTs in the ring segment S_a, with r_I ≤ r ≤ r_G and central angle ψ, are served by the UAV via cyclical multiple access.S_a is symmetric about the horizontal axis and has equal central angle ψ across arcs centered at the GBS.
  • 3) UAV-MT association:: The proposed association rule assigns MTs within S_a to the UAV at time t, thereby determining K_U(t).Cell-edge MTs are exclusively served by the UAV cyclically, non-cell-edge MTs by the GBS, and no handover is needed.
  • 3) UAV-MT association:: Each associated MT incurs access delay D_k ≜ T − τ_k, making the scheme most suitable for high-demand cell-edge MTs with less stringent delay requirements.MTs with stringent delay requirements can instead be served conventionally by the GBS.
  • 4) Lower bound of average throughput:: The lower-bound analysis uses d_max, the maximum UAV-MT horizontal distance in S_a, and b_min, the lower bound on normalized per-user bandwidth.The coverage condition requires r_c ≥ d_max.
  • 4) Lower bound of average throughput:: The instantaneous-rate lower bound R_U decreases with ψ because larger ψ increases d_max and decreases b_min.The UAV antenna gain G_U toward the coverage area also decreases with d_max.
  • 4) Lower bound of average throughput:: The common UAV-served average throughput R̄_U decreases with d_max, which increases with ψ, so ψ should be as small as possible subject to nonempty association.ψ cannot be arbitrarily small because |K_U(t)| may equal 0 at some time t.

B. GBS-MT Communication · 1) Power allocation:

The GBS serves non-cell-edge MTs inside an inner disk of radius rI, using average-gain-based slow channel inversion to equalize their average received SNR. The resulting distance-dependent power allocation is constrained by the GBS maximum transmit power and determines each MT’s achievable rate.

  • B. GBS-MT Communication: MTs inside the inner disk of radius rI are associated with the GBS, forming the non-cell-edge MT set KG.
  • B. GBS-MT Communication: The GBS-MT channel gain gk combines average gain ¯gk, determined by horizontal distance r, with small-scale fading ζk ∼Exp(1).
  • 1) Power allocation:: The GBS assumes equal power pG(r) for MTs at the same distance r, with r ≤rI.
  • 1) Power allocation:: The GBS uses “slow” channel inversion power control based on average channel gain ¯gk, avoiding costly instantaneous-channel estimation.
  • 1) Power allocation:: Power pG(r) is allocated so all MTs k ∈KG achieve the equal average receiver SNR ¯γ, with power inversely proportional to average channel gain.
  • 1) Power allocation:: The total transmit power for GBS-associated MTs must satisfy a constraint based on the GBS maximum transmit power PG.
  • 1) Power allocation:: The average SNR follows from the power-allocation relations, while the instantaneous achievable rate for MT k ∈KG is specified in bps/Hz.

2) Outage probability: … 1) Optimizing rU:

The section characterizes outage probability under small-scale fading, formulates joint optimization of throughput and network parameters, and develops a structured solution that optimizes the UAV trajectory radius before searching bandwidth allocation and user partitioning.

  • 2) Outage probability:: An outage occurs when the GBS-MT link cannot support the desired common throughput ¯ν under small-scale fading.The outage event is defined by the GBS-MT link’s inability to meet the common throughput requirement.
  • 2) Outage probability:: Pout,k = Pr{bG log2(1 + ¯γζk) < ¯ν} = Pr{ζk < (2¯ν/bG −1)/¯γ}.The expression gives the outage probability for each mobile terminal k.
  • 2) Outage probability:: Pout(ρ, rI, ¯ν) increases with bandwidth portion ρ, partitioning threshold rI, and common throughput ¯ν.This monotonicity follows from the defined function f(ρ, rI, ¯ν).
  • C. Problem Formulation: The optimization maximizes the common throughput ¯ν subject to the maximum GBS-MT outage constraint by jointly selecting ρ, rI, and rU.The variables represent bandwidth allocation, user partitioning distance, and UAV trajectory radius.
  • D. Proposed Solution: Problem (P1) is reduced to target-throughput subproblems (P2), with ¯ν updated by bisection according to whether the outage target is achievable.If P2’s optimum is no larger than ¯Pout, the target is feasible; otherwise, it is not achievable.
  • D. Proposed Solution: For fixed ρ and rI, (P3) optimizes rU to maximize the achievable UAV-MT common throughput while satisfying constraint (29).This decomposition exploits the independence of GBS-MT communication from rU.
  • D. Proposed Solution: The resulting (P4) remains non-convex in ρ and rI, so it is solved by inner bisection over ρ and outer one-dimensional search over rI.The search ranges are 0 < ρ < 1 and 0 ≤ rI ≤ rG.
  • 1) Optimizing rU:: Optimizing rU is equivalent to minimizing dmax = max(dA, dB), achieved by setting dA = dB when rI ≤ rU ≤ rG and ψ ≤ ψ0.When the geometric condition does not hold, the minimum dmax equals rG sin ψ.

2) Optimizing ρ and rI: … 1) Directional transmission:

The paper solves the orthogonal-sharing optimization by exploiting monotonicity and one-dimensional searches, then extends to spectrum reuse with directional antennas that enable concurrent UAV and GBS communications while jointly optimizing rU and rI.

  • 2) Optimizing ρ and rI:: For fixed ¯ν and rI, f(ρ, rI, ¯ν) increases with ρ while ¯ν −¯Rmax decreases with ρ.This monotonicity motivates selecting the smallest feasible ρ.
  • 2) Optimizing ρ and rI:: The algorithm searches 0 < ρ < 1 for the minimum feasible ρ, then performs a one-dimensional search over 0 ≤rI ≤rG.These searches further minimize f(ρ, rI, ¯ν) in (P4).
  • IV. SPECTRUM REUSE: In spectrum reuse, the common spectrum pool of total bandwidth W Hz is shared by both the GBS and UAV.The scheme is intended to improve spectrum efficiency when mutual interference is controlled.
  • IV. SPECTRUM REUSE: Directional/sectorized antennas at the UAV and GBS are jointly used to eliminate mutual interference and maximize throughput performance.This design supports interference-free concurrent communications between UAV-MT and GBS-MT links.
  • IV. SPECTRUM REUSE: Because ρ is unnecessary in spectrum reuse, the optimization jointly selects the UAV trajectory radius rU and user-partitioning threshold rI to maximize ¯ν.The objective is the minimum throughput ¯ν of all MTs.
  • 1) Directional transmission:: The GBS dynamically directs transmission toward shadowed sector Sb with central angle ΦG, non-overlapping the UAV association region Sa with central angle ψ.This directional arrangement causes no interference to UAV-MT communications.
  • 1) Directional transmission:: Non-cell-edge MTs in Sb remain associated with the GBS, which allocates equal effective normalized bandwidth bG(t) = 1/|KG(t)|.The passages give |KG(t)| = λr2IΦG/2 and bG(t) = 2/(λr2IΦG).
  • 1) Directional transmission:: Adaptive beamforming or sectorized-antenna on-off control lets the GBS adapt its direction as the UAV flies cyclically, producing cyclical multiple access with period T.Each GBS-served MT k ∈KG has access delay Dk = (1 −ΦG.

2) Power allocation: … D. Proposed Solution

The spectrum-reuse design allocates GBS power by slow channel inversion, yields lower outage probability than orthogonal sharing, and jointly optimizes user partitioning and UAV trajectory. Its solution uses bisection and interference avoidance to improve throughput, with added implementation complexity.

  • 2) Power allocation:: The GBS uses slow channel inversion, assigning equal power to MTs at the same distance so their average received SNR is equal.The total GBS transmit power remains constrained by its maximum transmit power PG.
  • 3) Outage probability:: The spectrum-reuse scheme minimizes outage probability at ρ = 0 and therefore outperforms orthogonal sharing under the same rI and ¯ν.Because outage probability increases with ρ, reuse achieves higher throughput under the same outage requirement.
  • B. UAV-MT Communication: With GBS interference eliminated, the UAV uses the whole bandwidth, and its common throughput depends on rI and rU.The UAV-MT communication follows the earlier model with ρ = 1.
  • C. Problem Formulation: The optimization maximizes common throughput subject to a GBS-MT outage constraint by jointly selecting user-partitioning threshold rI and UAV trajectory radius rU.Unlike the earlier formulation, bandwidth partitioning between the UAV and GBS is unnecessary.
  • D. Proposed Solution: For fixed rI, optimizing rU gives UAV throughput increasing with rI, whereas the maximum GBS-MT throughput decreases with rI.A bisection search over rI finds the optimum, with the GBS constraint active at equality.
  • D. Proposed Solution: The proposed solution requires adaptive directional GBS transmissions and cyclical multiple access, increasing implementation complexity.These mechanisms support interference avoidance while enabling concurrent GBS and UAV access to the common spectrum pool.
  • D. Proposed Solution: Interference avoidance lets the GBS and UAV concurrently access the common spectrum pool, further improving system throughput.The throughput improvement is stated as a consequence of concurrent communications under spectrum reuse.

E. Further Discussions … 3) Extension to Multiple Cells:

The proposed design remains applicable with variable UAV altitude and limited user-location information, while extensions to multiple UAVs and cells offer lower access delay, higher throughput, and interference-aware offloading directions for future work.

  • 1) Relaxation of fixed UAV altitude:: Optimizing UAV trajectory radius rU, bandwidth allocation portion ρ, and user-partitioning threshold rI jointly determines coverage radius rc for scheduled MTs.These parameters provide practical design guidelines while guaranteeing scheduled UAV-served MTs remain within the coverage area.
  • 1) Relaxation of fixed UAV altitude:: Variable UAV altitude causes no fundamental change because beamwidth ΦU can be adjusted to maintain fixed coverage radius rc = HU tan ΦU.The fixed-altitude assumption is therefore mainly for simplicity, and some altitude/beamwidth control error can be tolerated in practice.
  • 2) Requirement of User Location Information:: The optimization schemes require user-distribution statistics rather than exact ground-user locations, with served MTs identifiable over time using RSRP.Accurate location information may help but is not required for the proposed schemes.
  • 2) Requirement of User Location Information:: Under a homogeneous Poisson point process with density λ, a constant-speed circular trajectory along the cell edge shortens UAV communication distances and improves overall throughput.This model represents hotspot cells through larger user density λ.
  • 2) Requirement of User Location Information:: For known nonuniform “hotspots in hotspot,” optimizing trajectory and speed can further improve throughput by bringing the UAV closer or maintaining longer service duration.The UAV may fly closer to or hover above these concentrated user regions.
  • 3) Extension to Multiple Cells:: With multiple UAVs in one cell, equal separation along the designed trajectory helps reduce access delay and improve throughput; one UAV per cell extends the current results directly.These multi-cell extensions are motivated by the single-cell performance gain and remain future-work directions.
  • 3) Extension to Multiple Cells:: Multi-cell operation requires interference mitigation and collision avoidance, supported by directional UAV antennas and trajectories confined within cell boundaries.A single UAV could also serve multiple cells sequentially, while macro-cell offloading must account for existing macro- and micro-BS locations, partitioning, and spectrum sharing.
  • 3) Extension to Multiple Cells:: A UAV can offload an overloaded macro BS when few nearby small cells exist, but the corresponding trajectory, user partitioning, and spectrum sharing require further investigation.The design must reflect the locations and resource-sharing arrangements of existing macro and micro base stations.

4) Energy Constraint for UAVs: · V. NUMERICAL RESULTS · A. Performance Evalution of the Proposed Schemes

The paper addresses UAV endurance through multi-UAV rotation or automated battery swapping and evaluates optimized offloading designs against fixed-parameter and GBS-only benchmarks. Simulations show optimized orthogonal sharing and spectrum reuse improve spatial throughput and support substantially higher user densities.

  • 4) Energy Constraint for UAVs:: UAV endurance is limited by onboard energy, motivating multiple UAVs that rotate service while recharging or swapping batteries on the ground.Automated battery swap and recharge can let a single UAV conduct long-endurance missions.
  • 4) Energy Constraint for UAVs:: Fixed-wing UAVs typically offer much longer flight endurance than rotary-wing UAVs, making them suitable for the considered application.The paper also notes that Section V-B provides a quantitative energy-efficiency example.
  • 4) Energy Constraint for UAVs:: UAVs can be rapidly deployed for temporary hotspots when existing ground infrastructure cannot absorb sudden traffic surges and new infrastructure is costly or slow to install.The scheme is intended to provide high throughput to ground MTs temporarily.
  • V. NUMERICAL RESULTS: Numerical results validate the analysis using optimized and fixed design parameters, GBS-only comparisons, and conventional micro/small-cell edge-enhancement comparisons.Results are organized into evaluations of the proposed schemes and comparisons with conventional cell-edge enhancement.
  • A. Performance Evalution of the Proposed Schemes: The simulations average 100 independent homogeneous Poisson user-location realizations, with channel inversion power control simulated per realization.Average spatial throughputs are obtained separately for GBS- and UAV-served areas.
  • A. Performance Evalution of the Proposed Schemes: Analytical results match simulations, while optimized orthogonal sharing exceeds GBS-only spatial throughput for all PU values and fixed designs improve it when PU ≥10 dBm.The optimized design also outperforms the fixed-parameter case.
  • A. Performance Evalution of the Proposed Schemes: As PU increases, orthogonal sharing allocates more bandwidth to the UAV, whereas spectrum reuse serves more users through a decreasing optimal UAV-served-region ratio.Spectrum reuse significantly improves spatial throughput over corresponding orthogonal sharing, and joint optimization is essential for maximum throughput.

B. Illustrative Example of UAV Energy Efficiency

The section illustrates UAV energy-efficiency evaluation by measuring transmitted information per joule and incorporating propulsion power for circular fixed-wing flight. In an example orthogonal-sharing setup, the optimized trajectory achieves approximately 3.0 bps/Hz/km2 spatial throughput, with minimum propulsion power at 29.7 m/s.

  • Energy-efficiency evaluation: UAV energy efficiency is defined as transmitted information bits per unit energy consumed by the UAV.This definition accounts for the UAV’s dominant propulsion energy consumption.
  • Propulsion-power model: The fixed-wing UAV propulsion model accounts for instantaneous velocity and acceleration, including parasitic and induced power components.For constant-speed circular flight, the model uses velocity V and trajectory radius rU; c1 represents parasitic power and c2 induced power.
  • Propulsion-power model: For a given rU, an optimum speed V ∗ minimizes circular-trajectory propulsion power.The circular-trajectory UAV power consumption decreases with rU, while each fixed radius has a minimizing speed.
  • Numerical example: 3.0 bps/Hz/km2 is the obtained spatial throughput in the example orthogonal-sharing setup after optimizing the UAV trajectory radius.The setup uses ρ = 0.5, rI = 0.5rG, λ = 1000 MTs/km2, ψ = π/6, PU = 1 W, c1 = 9.26×10−4, and c2 = 2250.
  • Numerical example: 29.7 m/s is the optimum speed at which propulsion power is minimized in the example.The corresponding UAV propulsion power is obtained from the circular-trajectory power model.

C. Comparison with Micro-Cell Offload Scheme · VI. CONCLUSIONS

The optimized UAV offloading scheme outperforms micro-cell offloading for all tested micro-cell counts, while the proposed hybrid architecture jointly optimizes spectrum allocation, user partitioning, and UAV trajectory. Spectrum reuse with adaptive directional transmissions further improves spatial throughput, and the single-UAV design saves infrastructure cost, although extensions to multiple UAVs or cells remain challenging.

  • C. Comparison with Micro-Cell Offload Scheme: The benchmark deploys M uniformly placed micro-cell BSs at the macro-cell edge, serving overlapping-region MTs through their nearest micro BS.MTs outside micro-cell coverage remain associated with the GBS.
  • C. Comparison with Micro-Cell Offload Scheme: The micro-cell scheme allocates ρmicro of total bandwidth W equally among M micro BSs and shares each BS’s bandwidth equally among its associated MTs.The GBS and micro BSs use orthogonal spectrum sharing.
  • C. Comparison with Micro-Cell Offload Scheme: The micro-cell benchmark optimizes ρmicro, dmicro, and rmicro to maximize average minimum throughput and the resulting spatial throughput.The minimum throughput is ν = min{νG, νmicro}, averaged across N = 20 HPPP user-location realizations.
  • C. Comparison with Micro-Cell Offload Scheme: The micro-cell scheme’s spatial throughput increases with the number of micro cells as optimized placements move closer to the cell edge.This placement improves offloading relative to the GBS-only benchmark.
  • C. Comparison with Micro-Cell Offload Scheme: A single optimized UAV/mobile BS significantly outperforms the micro-cell offloading scheme for every tested M.The gain is attributed to UAV mobility and generally better LoS communication links to served ground MTs.
  • VI. CONCLUSIONS: The proposed hybrid architecture maximizes common MT throughput by jointly optimizing spectrum allocation, user partitioning, and UAV trajectory design.The study first considers orthogonal spectrum sharing between the UAV and GBS, then extends to spectrum reuse.
  • VI. CONCLUSIONS: Spectrum reuse shares the common spectrum pool between the GBS and UAV while suppressing mutual interference through adaptive directional transmissions, further improving spatial throughput.The hybrid design significantly improves throughput over the conventional GBS-only system.
  • VI. CONCLUSIONS: The optimized one-UAV scheme significantly outperforms multiple micro/small cells in throughput while saving infrastructure cost.Extending the work to multiple UAVs and/or multiple cells remains challenging and warrants investigation.
Loading 1705.09024v4…