PaperlayerView exact revision on arXiv

Source-linked AI summary

Mobile Unmanned Aerial Vehicles (UAVs) for Energy-Efficient Internet of Things Communications

Mohammad Mozaffari, Walid Saad, Mehdi Bennis, Merouane Debbah

arXiv:1703.05401v2cs.IT

TL;DR

The paper asks how multiple UAV aerial base stations can reliably collect uplink data from time-varying IoT devices while minimizing device transmit power. It jointly optimizes 3D placement, mobility, association, and power control, including update timing and minimum-energy trajectories. Compared with stationary aerial base stations, the approach reduces device transmit power and improves reliability, while exposing a tradeoff with UAV mobility and update frequency.

  • Problem

    The paper addresses reliable, energy-efficient uplink data collection from IoT devices by jointly considering multiple UAVs’ deployment, mobility, association, and uplink power control.

  • Method

    The framework optimizes UAV locations and device associations for active-device snapshots, then derives activation-based update times and minimum-energy 3D UAV trajectories for time-varying networks.

  • Results

    The approach reduces total IoT transmit power and improves reliability relative to stationary aerial base stations.

  • Takeaways & Limitations

    More frequent location updates can lower IoT transmit power but increase UAV mobility and energy consumption.

Abstract

from arXiv · show

In this paper, the efficient deployment and mobility of multiple unmanned aerial vehicles (UAVs), used as aerial base stations to collect data from ground Internet of Things (IoT) devices, is investigated. In particular, to enable reliable uplink communications for IoT devices with a minimum total transmit power, a novel framework is proposed for jointly optimizing the three-dimensional (3D) placement and mobility of the UAVs, device-UAV association, and uplink power control. First, given the locations of active IoT devices at each time instant, the optimal UAVs' locations and associations are determined. Next, to dynamically serve the IoT devices in a time-varying network, the optimal mobility patterns of the UAVs are analyzed. To this end, based on the activation process of the IoT devices, the time instances at which the UAVs must update their locations are derived. Moreover, the optimal 3D trajectory of each UAV is obtained in a way that the total energy used for the mobility of the UAVs is minimized while serving the IoT devices. Simulation results show that, using the proposed approach, the total transmit power of the IoT devices is reduced by 45% compared to a case in which stationary aerial base stations are deployed. In addition, the proposed approach can yield a maximum of 28% enhanced system reliability compared to the stationary case. The results also reveal an inherent tradeoff between the number of update times, the mobility of the UAVs, and the transmit power of the IoT devices. In essence, a higher number of updates can lead to lower transmit powers for the IoT devices at the cost of an increased mobility for the UAVs.

I. INTRODUCTION

The paper addresses energy-efficient and reliable uplink data collection from time-varying IoT devices by jointly optimizing multiple UAVs’ 3D deployment, mobility, association, and power control. It develops a two-step framework for snapshot deployment and dynamic movement, reporting lower transmit power and a tradeoff between update frequency, UAV mobility, and device power.

  • Motivation: UAVs serve as mobile aerial base stations that collect data from energy-constrained IoT devices spread across areas with limited terrestrial infrastructure.Their mobility lets them move toward devices and support communications beyond ordinary transmitter ranges.
  • Research gap: Prior studies did not jointly optimize UAV deployment and mobility, device association, and uplink power control for reliable, energy-efficient IoT communications.Earlier work also included settings with a single UAV or static sensor networks without optimal UAV deployment.
  • Framework: The proposed framework jointly determines multiple UAVs’ 3D locations, device associations, and uplink transmit powers under device-specific SINR constraints.For a fixed snapshot, deployment and association are optimized through two iteratively solved subproblems, including a convex reformulation of location optimization.
  • Framework: For time-varying device activity, the framework derives UAV update times and minimum-mobility 3D trajectories while continuing to serve the IoT devices.The update times follow the activation process, and the trajectories minimize total UAV movement during location updates.
  • Results: The proposed approach reduces IoT transmit power compared with stationary aerial base stations and reveals a tradeoff among update count, UAV mobility, and device transmit power.More updates lead to lower IoT transmit powers at the cost of higher UAV energy consumption.

II. SYSTEM MODEL AND PROBLEM FORMULATION

The paper models a centralized UAV-assisted IoT network in which UAV locations, device associations, and transmit powers are jointly selected over time to serve changing active-device sets. The formulation uses a ground-to-air channel model based on probabilistic LoS/NLoS propagation and average channel gain.

  • II. SYSTEM MODEL AND PROBLEM FORMULATION: The system contains L IoT devices and K rotary-wing UAVs that collect data through uplink FDMA over R orthogonal channels.Each UAV has a maximum energy budget, and successful service requires the device’s uplink SINR to exceed a threshold.
  • II. SYSTEM MODEL AND PROBLEM FORMULATION: A centralized cloud center knows device and UAV locations and determines UAV locations, device-UAV associations, and device transmit powers.Active devices are assumed to be known at the beginning of each service slot.
  • II. SYSTEM MODEL AND PROBLEM FORMULATION: At each update time, UAV positions and associations are optimized for the currently active devices, which are then served until the next update.Update times are design parameters, and UAV trajectories consist of stop locations associated with these updates.
  • II. SYSTEM MODEL AND PROBLEM FORMULATION: The optimization minimizes total transmit power while satisfying every active device’s SINR requirement and accounts for mobility, activity changes, and update costs.Continuous updating may be infeasible because it can reduce reliability, increase UAV energy consumption, and require complex real-time optimization.
  • II. SYSTEM MODEL AND PROBLEM FORMULATION: The proposed framework first solves placement, association, and power-control decisions, then optimizes UAV mobility for dynamically changing active devices.The framework also introduces the ground-to-air channel and IoT activation models needed for these decisions.
  • A. Ground-to-Air Path Loss Model: The ground-to-air model uses elevation angle, altitude, environment, and distance to characterize LoS probability and average path loss.Averaging LoS and NLoS links yields an average channel gain that makes SINR modeling more tractable.
  • A. Ground-to-Air Path Loss Model: Because exact obstacle information is unavailable, the model retains LoS/NLoS randomness and uses average path loss for desired and interfering links.The environment may be rural, urban, or dense urban, with corresponding propagation parameters.

B. IoT Device Activation Model

IoT activity may be random or periodic, so UAV deployment and update timing must adapt to changing active-device patterns. The proposed solution formulates and iteratively solves the coupled association, placement, and power-control decisions.

  • B. IoT Device Activation Model: IoT devices may generate bursty traffic through random activations or report periodically through deterministic, known activation patterns.Random activity is modeled with a beta distribution, while periodic devices activate every τ_i seconds.
  • B. IoT Device Activation Model: The optimal UAV locations and update times depend on device activation processes, requiring deployment that adapts to device activity.The framework considers randomly activated devices, including short-period concurrent transmissions from many devices.
  • C. Channel Assignment Strategy: Devices are grouped by proximity for channel assignment, and nearby devices are assigned different orthogonal channels to mitigate interference.The paper adopts constrained K-mean clustering for this assignment when R ≤ L_n.
  • C. Channel Assignment Strategy: At each update time, the formulation jointly represents active-device transmit powers, 3D UAV locations, and device-UAV associations under SINR and maximum-power constraints.The SINR target γ must be achieved by all active devices, while P_max limits each device’s transmit power.
  • C. Channel Assignment Strategy: UAV placement affects channel gains and transmit powers, while associations and interference also depend on placement and the powers of other devices.These mutual dependencies make the original optimization highly nonlinear and non-convex.
  • C. Channel Assignment Strategy: The proposed algorithm decomposes the original problem into alternating association-and-power and UAV-location subproblems.With fixed locations it optimizes association and power; with fixed association it updates UAV locations and powers until convergence.
  • C. Channel Assignment Strategy: Each iteration decreases total device transmit power, and the procedure converges after jointly updating 3D locations, associations, and transmit powers.Figure 2 summarizes the main solution steps.

III. UAV DEPLOYMENT AND DEVICE ASSOCIATION WITH POWER CONTROL

The deployment problem is solved by alternating between optimal association and power control for fixed UAV positions and suboptimal UAV repositioning for fixed associations. The resulting placement must satisfy SINR feasibility while balancing device power and UAV geometry.

  • III. UAV DEPLOYMENT AND DEVICE ASSOCIATION WITH POWER CONTROL: Given UAV locations, the first subproblem finds device associations and transmit powers that satisfy all active-device SINR requirements with minimum total power.The association and power solution is then used to optimize UAV locations in the second subproblem.
  • III. UAV DEPLOYMENT AND DEVICE ASSOCIATION WITH POWER CONTROL: The two subproblems are solved iteratively because UAV locations and device associations are mutually dependent.The resulting computations produce 3D UAV locations, device associations, and device transmit powers.
  • III. UAV DEPLOYMENT AND DEVICE ASSOCIATION WITH POWER CONTROL: When active devices do not exceed orthogonal channels, interference is absent; otherwise, the association and power-control problem includes uplink interference.The framework analyzes interference and interference-free cases separately.
  • III. UAV DEPLOYMENT AND DEVICE ASSOCIATION WITH POWER CONTROL: Altitude feasibility depends on device-UAV distance, SINR requirements, and device maximum transmit power.The derived bounds require the elevation angle to exceed a threshold, while the maximum altitude depends on P_max.
  • III. UAV DEPLOYMENT AND DEVICE ASSOCIATION WITH POWER CONTROL: For fixed UAV locations, the iterative association and power-control algorithm converges to the global optimum when every device’s SINR equals γ.It repeatedly selects the best UAV and updates each device’s power toward its SINR target.
  • III. UAV DEPLOYMENT AND DEVICE ASSOCIATION WITH POWER CONTROL: The algorithm outputs the optimal transmit-power vector and device-association vector for the given UAV locations.Steps 4–7 are repeated for all devices until the power vector converges.

2) Interference-free scenario:

In the interference-free case, association reduces to assigning devices to UAVs using path-loss-based costs subject to maximum-power feasibility. The resulting assignment can be solved efficiently as a classical assignment problem.

  • 2) Interference-free scenario:: When L_n ≤ R, each active device can use an orthogonal channel, so interference is zero and transmit power depends only on its serving-UAV channel gain.For fixed UAV locations, the minimum power is P_i = γσ^2 L̄_ij.
  • 2) Interference-free scenario:: The interference-free association problem minimizes total path-loss-based assignment cost using binary device-UAV assignment variables.The average path loss L̄_ij is known from device and UAV locations.
  • 2) Interference-free scenario:: The constrained integer linear program is transformed into a standard assignment problem solvable in polynomial time by the Hungarian method.The stated time complexity is O((L_nK)^3).
  • 2) Interference-free scenario:: The transformed assignment approach is motivated by the potentially high number of IoT devices, whereas generic integer-programming methods may become inefficient as problem size grows.In the absence of interference, the interference and interference-free subproblems have the same solution.
  • 2) Interference-free scenario:: Maximum-power feasibility is enforced by assigning infinite cost to device-UAV pairs whose required power exceeds P_max.This prevents assignments that cannot satisfy the device power constraint.

B. Optimal Locations of the UAVs

The UAV-location problem is non-linear and non-convex because device transmit powers and UAV positions are mutually dependent. The proposed iterative approach optimizes UAVs one by one, updating associated-device powers until no further reduction is possible.

  • The approach determines sub-optimal 3D UAV locations while reducing device transmit power after serving-UAV location updates.The method separately updates UAV locations and associated-device powers within the joint optimization process.
  • Device transmit powers and UAV locations are mutually dependent, making the optimization highly non-linear and non-convex.The coupling arises because UAV locations affect channel gains and transmit powers, while interference can also depend on UAV placement.
  • The cloud optimizes each UAV separately using fixed powers for non-associated devices and then updates associated-device powers after relocating its serving UAV.This process uses known associations and transmit powers from the preceding subproblem.
  • The UAV-location and power-update process repeats across UAVs until changing their locations cannot further reduce total device transmit power.At each step, the location is selected to minimize the transmit power of the UAV’s associated devices.
  • Sequential quadratic programming approximates the objective quadratically, linearizes constraints, and solves successive constrained quadratic subproblems.An interior-point method is used for the resulting large, sparse Hessian structure when many active devices create numerous constraints.

2) Interference-free scenario:

In the interference-free scenario, a quadratic approximation makes the UAV-location optimization tractable. The resulting procedure obtains sub-optimal 3D locations through convex optimization and iterative association-location updates that monotonically reduce transmit power.

  • In the absence of interference, the location optimization admits a tractable formulation based on a quadratic approximation.The approximation is constructed from bounds on the distance-dependent propagation function.
  • The quadratic approximation produces less than 4% error in the objective function for different UAV altitudes.This error assessment is reported for the parameters used in Table I.
  • The resulting quadratically constrained quadratic program is convex because its relevant matrices are positive semidefinite.The solution is derived through the Lagrange dual formulation and Theorem 1.
  • Iterating device association and UAV-location optimization reduces total transmit power monotonically until the solution converges after several iterations.The proposed method is suboptimal but reported as reasonably accurate and significantly faster than brute-force global search.

IV. UPDATE TIMES AND MOBILITY OF UAVS

The mobility framework determines when UAVs should update their locations as active IoT devices change over time. It also seeks trajectories that maintain reliable uplink service while minimizing UAV mobility energy under discrete stop-and-serve operation.

  • The framework finds UAV update times and trajectories to provide reliable uplink communications as the active-device set changes.UAV trajectories consist of stop locations at which associated ground devices are served.
  • UAV trajectories and update times depend on the IoT-device activation process.The UAVs stop to serve devices and then relocate rather than moving continuously.
  • The mobility objective is to minimize total UAV energy consumption so the UAVs remain operational longer.Mobility is limited by the UAVs’ energy constraints.
  • The framework uses update times t_n over [0,T] and jointly optimizes UAV locations, device associations, and transmit powers at each update.For tractability, devices are assumed synchronized at t = 0, although the joint optimization does not depend on this assumption.

A. Update Time Analysis

Update times are derived from the device activation process and the number of devices that can be served. More frequent updates shorten service intervals and can reduce device transmit power, but increase UAV mobility and energy consumption.

  • The proposed framework determines UAV update times from IoT-device activation patterns and the number of devices served at each update.The analysis considers probabilistic and periodic activation models.
  • A higher number of updates shortens the intervals between consecutive updates and reduces the number of active devices in each interval.Fewer simultaneously active devices experience lower mutual interference during uplink transmission.
  • Lower interference enables active IoT devices to use lower transmit power to meet the SINR constraint.The reported relationship connects update frequency with device-side power consumption.
  • More updates require greater UAV mobility and higher UAV energy consumption.This creates a tradeoff between update frequency, UAV mobility, and IoT transmit power.
  • Knowing the exact number of active devices enables deterministic and efficient update-time selection based on system requirements.Proposition 2 provides the exact number of devices to be served at each update time.

2) Probabilistic IoT activation:

The paper models probabilistic IoT-device activations and derives UAV update times from the expected number of devices requiring service. Update timing depends on activation patterns, device count, prior update times, and system constraints such as interference and available channels.

  • The probabilistic activation model assumes each IoT device becomes active during [0, T] according to a beta distribution.
  • Theorem 2 gives update times for a target average number of active devices using the regularized incomplete beta function and its inverse.
  • The probability that a device must be served at update time t_n is determined by whether it became active during [t_{n−1}, t_n).
  • Update times depend on the number of devices, their activation distribution, and the preceding update time because service demand depends on t_n − t_{n−1}.
  • The update schedule can balance interference avoidance and resource utilization by keeping the average number of active devices below the number of orthogonal channels.

B. UAVs’ Mobility

The UAV mobility procedure maps UAVs between successive optimized stop locations while minimizing transportation energy under per-UAV energy limits. It models vertical and horizontal flight consumption, solves the resulting assignment problem, and then derives energy-efficient trajectories.

  • The method obtains each UAV’s optimal trajectory between destinations so the UAVs remain operational while providing reliable and energy-efficient uplink transmissions.
  • The mobility optimization maps UAV locations at consecutive update times to minimize transportation energy while respecting each UAV’s remaining energy.
  • The assignment matrix Z indicates which UAV moves from each prior location to each new destination, while E_kl gives the corresponding movement energy.
  • Movement energy accounts for flight distance, duration, vertical power, and horizontal power, with velocity components determined by the altitude difference.
  • The integer linear mobility problem is transformed into a standard assignment problem solved by the Hungarian method with complexity O(|I_n|^3).

V. SIMULATION RESULTS AND ANALYSIS

Simulations evaluate reliability, transmit power, activation timing, and UAV mobility under the proposed dynamic deployment. Compared with stationary aerial base stations, the approach reduces transmit power and improves reliability, while more frequent updates trade lower device interference and power for greater UAV mobility and energy use.

  • Reliability: 28% maximum improvement in system reliability is achieved over stationary aerial base stations.With P_max increasing from 40 mW to 100 W, reliability rises from 0.58 to 0.82 for the proposed approach, versus 0.3 to 0.72 for the stationary case.
  • Transmit power: 45% average reduction in total IoT-device transmit power is achieved compared with the stationary case.The power-reduction gain is larger with fewer UAVs; for example, the gain with 5 UAVs is around seven times that with 10 UAVs.
  • Transmit power: 80% average transmit-power reduction is achieved in the interference-free scenario compared with stationary deployment.With 5 UAVs, reliable uplink communications require a total device transmit power of 70 mW.
  • Update tradeoff: More frequent updates increase UAV mobility and energy consumption, creating a tradeoff between device transmit power and UAV movement cost.The UAVs must move farther on average as the geographical service area grows.

VI. CONCLUSION

The paper develops a framework for jointly optimizing UAV placement, device association, uplink power control, and mobility in time-varying IoT networks. It minimizes device transmit power under SINR constraints while deriving UAV update times and energy-efficient trajectories, revealing a tradeoff between update frequency, UAV mobility, and transmit power.

  • The framework jointly optimizes UAV locations, device association, and uplink power control to minimize total device transmit power under SINR constraints.
  • It determines UAV update time instances from the IoT devices’ activation process in a time-varying network.
  • The method obtains UAV trajectories that dynamically serve IoT devices with minimum mobility energy consumption.
  • The results show that intelligently moving and deploying UAVs significantly decreases total device transmit power compared with pre-deployed stationary aerial base stations.
  • A fundamental tradeoff links the number of UAV location updates, UAV mobility, and device transmit power.
Loading 1703.05401v2…