pith. sign in

arxiv: 2604.27640 · v1 · submitted 2026-04-30 · 💻 cs.NI · cs.SY· eess.SY

Multi-Connectivity for UAVs: A Measurement Study of Integrating Cellular, Aerial Mesh, and LEO Satellite Links

Aygun Baltaci , Irshad A. Meer , Mustafa Ozger , Cicek Cavdar , Dominic Schupke This is my paper

Pith reviewed 2026-05-07 05:42 UTC · model grok-4.3

classification 💻 cs.NI cs.SYeess.SY
keywords UAVmulti-connectivityMPTCPLEO satelliteaerial meshcellularreal-time deliverymeasurement study
0
0 comments X

The pith

In UAV networks combining aerial mesh, cellular, and LEO satellite links, preserving end-to-end connectivity fails to guarantee timely real-time delivery because of large RTT differences.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper runs UAV flight experiments that combine UAV-to-UAV aerial mesh, private cellular, and LEO satellite links, then aggregates them with MPTCP as a representative multipath transport. It shows that this setup keeps the connection alive during individual link outages, yet the resulting packet reordering and receiver buffering from RTT heterogeneity produce bursty, delayed delivery that violates real-time constraints, even when total capacity meets the load. Sender buffering appears when any link cannot carry its share. The authors therefore distinguish connectivity continuity from service continuity and conclude that the first does not imply the second for delay-sensitive UAV traffic. A reader cares because many emerging UAV uses, such as live video or command-and-control, need both robustness and bounded latency, not just an open socket.

Core claim

Using MPTCP over an integrated UAV network of aerial mesh, private cellular, and LEO satellite links maintains end-to-end connectivity under severe link outages. However, large RTT heterogeneity across the paths amplifies packet reordering, which forces substantial receiver-side buffering and bursty delivery. When any link lacks sufficient capacity for the offered load, sender-side buffering also appears. These effects cause real-time streaming to miss its delay bounds even in cases where aggregate capacity is adequate. The measurements therefore establish that connectivity continuity is necessary but not sufficient for service continuity in heterogeneous multi-technology UAV networks.

What carries the argument

MPTCP aggregation over the three heterogeneous links, together with the explicit distinction between connectivity continuity and service continuity, which isolates buffering and reordering as the mechanisms that break real-time constraints.

If this is right

  • Multipath transport for UAVs must incorporate explicit delay or reordering awareness rather than optimizing only for connectivity or aggregate throughput.
  • Real-time UAV applications will experience latency violations whenever path RTTs differ substantially, regardless of whether total capacity is sufficient.
  • Sender buffering will appear and compound delays as soon as any single path cannot sustain its share of the offered load.
  • Service continuity metrics that track buffering and delivery jitter are needed in addition to simple connectivity checks.
  • Future multi-connectivity designs should include path selection or scheduling that respects delay bounds in addition to outage resilience.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Protocols that tolerate or conceal reordering, such as certain QUIC variants or application-level forward error correction, may be more suitable than strict in-order MPTCP for UAV real-time traffic.
  • The same RTT-heterogeneity problem is likely to appear in other mobile platforms that mix terrestrial, aerial, and satellite links, such as connected cars or maritime vessels.
  • A lightweight scheduler that drops or redirects packets from the slowest path when reordering exceeds a threshold could be tested as a direct extension of the observed buffering behavior.

Load-bearing premise

The specific flight paths, link qualities, and traffic loads tested are representative of real-world UAV real-time services and generalize beyond the private cellular plus mesh plus LEO setup used in the experiments.

What would settle it

A repeat of the flight trials in which the same UAV and load are run but the three paths are tuned or selected so that RTT variance stays below a few tens of milliseconds; if real-time video then meets its delay bounds without extra buffering, the claim that heterogeneity alone drives the service failures would be falsified.

Figures

Figures reproduced from arXiv: 2604.27640 by Aygun Baltaci, Cicek Cavdar, Dominic Schupke, Irshad A. Meer, Mustafa Ozger.

Figure 1
Figure 1. Figure 1: Multi-technology link handover demonstration setup at the Airbus test site in Ottobrunn/Munich, Germany. view at source ↗
Figure 2
Figure 2. Figure 2: Flight measurement setup. the heterogeneous links is terminated at a common gateway and forwarded to the application endpoint. B. Transport Layer Mechanism and Traffic Characteristics Traffic aggregation across heterogeneous links is performed using a lossless, in-order multipath transport mechanism, which is multipath Transmission Control Protocol (MPTCP) [6]. This design reflects current practice in mult… view at source ↗
Figure 3
Figure 3. Figure 3: Empirical cumulative distribution function (CDF) of round-trip times view at source ↗
Figure 6
Figure 6. Figure 6: Instantaneous achieved data rate over time during the UAV flight. view at source ↗
Figure 5
Figure 5. Figure 5: Dominant communication link over time, defined as the link view at source ↗
read the original abstract

Future uncrewed aerial vehicle (UAV) systems increasingly combine heterogeneous communication technologies, such as low-latency aerial mesh, terrestrial cellular, and satellite links, to improve robustness and coverage. Multipath transport is a natural mechanism for aggregating these links, yet its ability to support real-time UAV services in highly heterogeneous environments remains insufficiently characterized. We present a measurement-driven study based on UAV flight experiments in an integrated network comprising UAV-to-UAV aerial mesh, private cellular, and low Earth orbit (LEO) satellite connectivity. Using Multipath TCP (MPTCP) as a representative lossless, in-order multipath transport framework, we find that aggregation can preserve end-to-end connectivity under severe link outages. However, large round-trip time (RTT) heterogeneity amplifies packet reordering, leading to substantial receiver-side buffering and bursty delivery. In addition, when the available links do not provide sufficient capacity for the offered load, pronounced sender-side buffering emerges. These effects cause real-time streaming to violate delay constraints, including cases where aggregate capacity is sufficient. To interpret these results, we formalize the distinction between connectivity continuity and service continuity and show empirically that maintaining connectivity is necessary but not sufficient for timely real-time delivery in multi-technology UAV networks. The findings motivate multipath designs that explicitly account for delay constraints, rather than optimizing for connectivity alone.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The manuscript reports UAV flight experiments integrating aerial mesh, private cellular, and LEO satellite links with MPTCP. It shows that multipath aggregation preserves end-to-end connectivity during link outages, but large RTT heterogeneity induces packet reordering and receiver-side buffering while insufficient per-link capacity induces sender-side buffering; both effects produce delay violations for real-time streaming even when aggregate throughput meets the offered load. The authors formalize the distinction between connectivity continuity and service continuity and conclude that the former is necessary but not sufficient for timely real-time delivery in multi-technology UAV networks.

Significance. If the empirical results hold under representative conditions, the work supplies concrete evidence that standard lossless multipath transport is inadequate for delay-sensitive UAV services in heterogeneous environments. The flight-based measurements add practical value to UAV networking research and motivate delay-aware multipath designs rather than connectivity-centric ones.

major comments (3)
  1. [Experimental Methodology] The experimental methodology section provides insufficient detail on the number of flights, trial durations, exact offered loads for the real-time streaming traffic, and statistical treatment of results (e.g., variance or confidence intervals on buffering delays and delay-violation rates). Without these, it is impossible to judge whether the observed reordering and buffering effects are robust or sensitive to post-hoc selection of conditions.
  2. [Results and Analysis] The central claim that connectivity is necessary but not sufficient rests on the magnitude of RTT heterogeneity (LEO vs. mesh) and the chosen load patterns. The manuscript does not include sensitivity analysis or quantitative characterization of how varying RTT spreads or load levels affect reordering-induced buffering, leaving the generality of the insufficiency result unclear.
  3. [Discussion and Conclusions] The discussion of representativeness is limited. The specific private-cellular + aerial-mesh + LEO configuration and flight dynamics may not match operational UAV real-time services; the paper should explicitly compare the tested RTT distributions, capacity ratios, and traffic patterns against typical video/telemetry workloads to support extrapolation beyond the tested setup.
minor comments (2)
  1. [Figures] Figure captions and axis labels should explicitly state the number of aggregated runs or flights contributing to each plotted statistic.
  2. [Introduction] The formalization of connectivity vs. service continuity is introduced late; moving a concise definition to the introduction would improve readability.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive and detailed review. The comments on experimental rigor, generality, and representativeness are well taken. We have revised the manuscript to incorporate additional details, analysis, and discussion as described below, while preserving the core empirical focus of the measurement study.

read point-by-point responses

  1. Referee: [Experimental Methodology] The experimental methodology section provides insufficient detail on the number of flights, trial durations, exact offered loads for the real-time streaming traffic, and statistical treatment of results (e.g., variance or confidence intervals on buffering delays and delay-violation rates). Without these, it is impossible to judge whether the observed reordering and buffering effects are robust or sensitive to post-hoc selection of conditions.

    Authors: We agree that the original manuscript lacked sufficient quantitative detail on the experimental campaign. In the revised version, Section III has been expanded to report: 12 flights conducted over three days with a total of 18 trials; average trial duration of 14.7 minutes; exact offered loads of 2 Mbps and 4 Mbps constant-bit-rate real-time streams; and statistical treatment including means, standard deviations, and 95% confidence intervals for buffering delays and delay-violation rates, computed across all trials from the complete dataset. These additions allow direct assessment of robustness. revision: yes

  2. Referee: [Results and Analysis] The central claim that connectivity is necessary but not sufficient rests on the magnitude of RTT heterogeneity (LEO vs. mesh) and the chosen load patterns. The manuscript does not include sensitivity analysis or quantitative characterization of how varying RTT spreads or load levels affect reordering-induced buffering, leaving the generality of the insufficiency result unclear.

    Authors: The manuscript presents empirical results under the observed RTT heterogeneity (mesh 20-80 ms, LEO 300-800 ms) and load patterns that are characteristic of integrated UAV multi-connectivity. A full parametric sensitivity sweep across arbitrary RTT spreads is outside the scope of this measurement-driven study. However, the revised manuscript adds a new figure and quantitative analysis showing reordering buffer occupancy as a function of RTT difference at the tested loads, demonstrating that the observed buffering scales directly with the measured heterogeneity. We therefore maintain that the insufficiency result holds for the heterogeneous conditions examined, while acknowledging that broader modeling would be needed for exhaustive generality. revision: partial

  3. Referee: [Discussion and Conclusions] The discussion of representativeness is limited. The specific private-cellular + aerial-mesh + LEO configuration and flight dynamics may not match operational UAV real-time services; the paper should explicitly compare the tested RTT distributions, capacity ratios, and traffic patterns against typical video/telemetry workloads to support extrapolation beyond the tested setup.

    Authors: We have substantially expanded the Discussion section. The revised text now directly compares our measured RTT distributions (aerial mesh 15-60 ms, private cellular 30-120 ms, LEO 250-700 ms) against values reported for comparable UAV and LEO deployments in the literature. Capacity ratios (mesh up to 50 Mbps, cellular ~20 Mbps, LEO ~10 Mbps) are contrasted with typical UAV telemetry (0.1-1 Mbps) and real-time video (1-10 Mbps) workloads. The constant-bit-rate traffic patterns at loads matching or exceeding per-link capacity are aligned with real-time streaming requirements. While the private-cellular testbed and specific flight trajectories are particular to our setup, the degree of heterogeneity and outage behavior are representative of operational multi-technology UAV scenarios, supporting the stated conclusions. revision: yes

Circularity Check

0 steps flagged

No circularity: purely empirical measurement study with direct experimental grounding

full rationale

The paper is a measurement-driven study reporting UAV flight experiments with MPTCP over heterogeneous links (aerial mesh, private cellular, LEO satellite). The central claim—that connectivity continuity is necessary but not sufficient for service continuity—is supported directly by observed outcomes on reordering-induced receiver buffering, capacity-induced sender buffering, and resulting delay violations, even when aggregate throughput suffices. No mathematical derivations, fitted parameters, or predictions appear; the distinction between connectivity and service continuity is presented as an interpretive formalization of the empirical results rather than a self-referential definition. No self-citations are load-bearing in the provided text, and the study reports concrete measurements under explicitly stated flight conditions and loads without reducing any result to its own inputs by construction. This is a standard honest empirical paper whose findings stand or fall on the representativeness of the tested conditions, not on any circular chain.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The work is measurement-driven and introduces no free parameters, mathematical axioms, or invented entities. The primary unstated premise is that the tested scenarios and loads adequately represent broader UAV real-time use cases.

axioms (1)
  • domain assumption The experimental flight conditions, link qualities, and traffic loads are representative of practical UAV real-time services.
    The claims about delay violations and the necessity of service continuity rest on this representativeness assumption.

pith-pipeline@v0.9.0 · 5570 in / 1343 out tokens · 37365 ms · 2026-05-07T05:42:25.506854+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

6 extracted references · 6 canonical work pages

  1. [1]

    A survey of wireless networks for future aerial communications (facom),

    A. Baltaci, E. Dinc, M. Ozger, A. Alabbasi, C. Cavdar, and D. Schupke, “A survey of wireless networks for future aerial communications (facom),” IEEE Communications Surveys & Tutorials, vol. 23, no. 4, pp. 2833–2884, 2021

    work page 2021

  2. [2]

    Mobility management for cellular-connected uavs: Model-based versus learning- based approaches for service availability,

    I. A. Meer, M. Ozger, D. A. Schupke, and C. Cavdar, “Mobility management for cellular-connected uavs: Model-based versus learning- based approaches for service availability,”IEEE Transactions on Network and Service Management, vol. 21, no. 2, pp. 2125–2139, 2024

    work page 2024

  3. [3]

    Integration of satellite and aerial communications for heterogeneous flying vehicles,

    M. V ondra, M. Ozger, D. Schupke, and C. Cavdar, “Integration of satellite and aerial communications for heterogeneous flying vehicles,”IEEE network, vol. 32, no. 5, pp. 62–69, 2018

    work page 2018

  4. [4]

    Reliability and delay analysis of 3-dimensional networks with multi-connectivity: Satellite, haps, and cellular communications,

    F. Salehi, M. Ozger, and C. Cavdar, “Reliability and delay analysis of 3-dimensional networks with multi-connectivity: Satellite, haps, and cellular communications,”IEEE Transactions on Network and Service Management, vol. 21, no. 1, pp. 437–450, 2023

    work page 2023

  5. [5]

    Multipath transport analysis over cellular and LEO access for aerial vehicles,

    A. Baltaci, K. Chavali, M. Kosek, N. Mohan, D. A. Schupke, and J. Ott, “Multipath transport analysis over cellular and LEO access for aerial vehicles,”IEEE Access, vol. 11, pp. 118 490–118 511, 2023

    work page 2023

  6. [6]

    A. Ford, C. Raiciu, M. Handley, and O. Bonaventure,TCP Extensions for Multipath Operation With Multiple Addresses. document RFC 6824, Jan. 2013

    work page 2013