Starlink, as a representative low Earth orbit (LEO) satellite broadband system, makes high-bitrate video streaming possible in regions where terrestrial broadband is unavailable. However, its access links exhibit rapid throughput fluctuations caused by satellite mobility and handovers. Existing learned adaptive bitrate (ABR) algorithms can achieve high average quality of experience (QoE), yet high-bitrate Starlink streaming exposes severe session-level rebuffering that is not captured by average QoE alone. To address it, this paper proposes SafeSABR, a risk-calibrated learned ABR framework for Starlink networks. SafeSABR formulates Starlink ABR as a QoE--severe-risk tradeoff and follows a three-stage design: behavior-cloning pretraining learns a high-QoE ABR prior, risk-calibrated reinforcement learning (RL) fine-tuning reduces severe-tail action tendencies, and a runtime safety auditor uses safe-capacity lower bounds to check policy-requested bitrates before execution. Experiments on real Starlink traces compare SafeSABR with online, prediction-assisted, and learned ABR baselines. Compared with advanced methods, SafeSABR reduces severe-stall sessions from 22.8% to 7.2% and worst-5
Earth observation (EO) satellites in Low Earth Orbit (LEO) are collecting vast amounts of data, which are invaluable for applications such as monitoring forest fires. However, data downloading from EO satellites faces significant challenges due to the limited number of ground stations and the brief communication windows with them. Conversely, emerging LEO constellations like Starlink have enabled continuous connectivity and revolutionized access for ordinary users globally, who can connect via a simple satellite dish. In this paper, we study the feasibility of supporting EO satellites with Starlink satellite infrastructure and introduce a novel data delivery system, designated as "Starlink Space User" (SSU), for relaying data from observation satellites. SSU treats EO satellites as space users of Starlink, facilitating efficient data transfer to Earth. At the core of SSU is a novel class of algorithms designed for link and PoP selection, as well as system scheduling optimization, that operate effectively atop Starlink's proprietary infrastructure. We assess the performance of SSU using trace-driven simulations alongside real-world Starlink performance measurements. Our results demo
Global Navigation Satellite Systems (GNSS) underpin positioning, navigation, and timing (PNT), yet their low-power signals are easily blocked or disrupted, leaving gaps in PNT availability in contested environments (e.g. maritime settings) where interference, spoofing, or denial can occur. A key practical need is an independent, ubiquitous aiding signal that can be tracked passively and fused with inertial sensing to sustain full navigation-state estimation without dedicated or cooperative infrastructure. This paper presents an end-to-end LEO-aided hybrid framework that fuses GPS, Starlink downlink beacons, and an inertial measurement unit (IMU) in a 9D (3D position, 3D velocity, and 3D attitude) PNT system using an extended Kalman filter (EKF). We (i) extract Doppler-rate from Starlink downlink beacon tones by associating measurements with satellite IDs, (ii) benchmark beacon Doppler-rate against OFDM-derived range observables under a common processing/estimation pipeline, and (iii) integrate the resulting observable into inertial navigation. We evaluate GPS/IMU, Starlink/IMU, and GPS-Starlink-IMU using Fisher-information predictions, Monte Carlo simulations, and hardware measurem
Starlink has rapidly emerged as the world's largest satellite constellation and the de facto reference system for low Earth orbit (LEO) networking research. Existing literature predominantly models Starlink as a static, symmetric, and fully deployed structure with uniformly distributed satellites. However, we reveal that Starlink's actual deployment, orbital configurations, and operational dynamics fundamentally deviate from these idealized assumptions. Leveraging satellite observation data spanning 2019 to 2025, we demonstrate that the constellation is highly dynamic across multiple temporal and spatial scales. Macroscopically, Starlink comprises multiple orbital shells undergoing continuous active deployment and reconfiguration. Microscopically, individual satellites exhibit high mobility, frequently executing collision-avoidance maneuvers, altitude adjustments, and intra-orbital relocations. We discover that while the majority of satellites form a relatively stable structure with near-uniform spacing, other satellites tend to cluster as twins or triads as in-orbit backups. Furthermore, empirical survival analysis indicates an operational lifespan of 4-6 years and an average dail
We identify and characterize dedicated pilot symbols and other predictable elements embedded within the Starlink Ku-band downlink waveform. Exploitation of these predictable elements enables precise opportunistic positioning, navigation, and timing using compact, low-gain receivers by maximizing the signal processing gain available for signal acquisition and time-of-arrival (TOA) estimation. We develop an acquisition and demodulation framework to decode Starlink frames and disclose the explicit sequences of the edge pilots -- bands of 4QAM symbols located at both edges of each Starlink channel that apparently repeat identically across all frames, beams, channels, and satellites. We further reveal that the great majority of QPSK-modulated symbols do not carry high-entropy user data but instead follow a regular tessellated structure superimposed on a constant reference template. We demonstrate that exploiting frame-level predictable elements yields a processing gain of approximately 48 dB, thereby enabling low-cost, compact receivers to extract precise TOA measurements even from low-SNR Starlink side beams.
In the last few years, considerable research efforts have focused on measuring and improving Starlink network performance, especially for user terminals (UTs) in stationary scenarios. However, the performance of Starlink networks in mobility settings, particularly with frequent changes in the UT's orientation, and the impact of environmental factors, such as transient obstructions, has not been thoroughly studied, leaving gaps in understanding the causes of performance degradation. Recently, researchers have started identifying the communicating satellites to evaluate satellite selection strategies and the impact on network performance. However, existing Starlink satellite identification methods only work in stationary, obstruction-free scenarios, as they do not account for UT mobility, obstructions or detect dynamic beam switching events. In this paper, we reveal that the UT can perform multiple dynamic beam switching attempts to connect to different satellites when the UT-satellite link is degraded. This degradation can occur either due to the loss of line-of-sight (LoS) from changes in the FOV or obstructions, or due to poor signal quality, extending UT-satellite handovers beyon
We present the largest survey to date characterising intended and unintended emission from Starlink satellites across the SKA-Low frequency range. This survey analyses ~76 million full sky images captured over ~29 days of observing with an SKA-Low prototype station - the Engineering Development Array 2 - at the site of SKA-Low. We report 112,534 individual detections of 1,806 unique Starlink satellites, some emitting broadband emission and others narrowband emission. Our analysis compares observations across different models of Starlink satellite, with 76% of all v2-mini Ku and 71% of all v2-mini Direct to Cell satellites identified. It is shown that in the worst cases, some datasets have a detectable Starlink satellite in ~30% of all images acquired. Emission from Starlink satellites is detected in primary and secondary frequency ranges protected by the International Telecommunication Union, with 13 satellites identified between 73.00 - 74.60 MHz and 703 identified between 150.05 - 153.00 MHz. We also detect the reflections of terrestrial FM radio off different models of Starlink satellites at 99.70 MHz. The polarisation of the broadband emission shows the flux density of two orth
The evolution of 6th generation (6G) networks increasingly relies on satellite-based Non-Terrestrial Networks (NTNs) to extend broadband connectivity to remote and unserved regions, and to support public safety. In this paper we compare two representative and conceptually different satellite constellation architectures, namely Starlink and IRIS 2. Starlink is a commercial private Internet constellation by SpaceX, based on dense Low Earth Orbit (LEO) satellites. It is primarily designed to deliver high-capacity broadband services for civil applications, with performance targets comparable to those of terrestrial networks. In contrast, IRIS 2 is a planned public initiative to be deployed by the European Union, based on a multi-layer combination of LEO, Medium Earth Orbit (MEO), and Geo-stationary Earth Orbit (GEO) satellites. It is primarily designed to provide a secure, resilient, and sovereign infrastructure for government and critical communications. After describing the main technical characteristics of Starlink and IRIS 2, we run a comprehensive simulation campaign to evaluate the design tradeoffs between the two. Specifically, we evaluate the per-cell and per-user achievable ca
As a representative low Earth orbit (LEO) broadband system, Starlink exhibits highly variable access throughput, making short-term forecasting essential for network resource management. Existing forecasting methods mainly optimize symmetric point-prediction metrics such as MAE and RMSE, but they do not explicitly control the asymmetric risk of overestimating future throughput, which can cause over-admission, bandwidth overbooking, and service violations. This paper formulates Starlink throughput prediction as a risk-budgeted safe forecasting problem, where the predictor must satisfy a prescribed overestimation budget while maintaining competitive accuracy. We propose Budget-Guided Coarse-to-Fine Quantile Selection (BG-CFQS), a data-driven framework that trains a family of lower-quantile predictors, locates the quantile boundary satisfying the risk budget, and refines the boundary region to select the most accurate feasible predictor. Experiments on three real-world Starlink throughput datasets show that BG-CFQS satisfies the risk budget on all datasets and achieves the lowest average MAE, mean positive error, and tail positive error among budget-feasible methods. In high-risk and s
Satellite communications have emerged as one of the most feasible solutions to provide global wireless coverage and connect the unconnected. Starlink dominates the market with over 7,000 operational satellites in low Earth orbit (LEO) and offers global high-speed and low-latency Internet service for stationary and mobile use cases, including in-motion connectivity for vehicles, vessels, and aircraft. Starlink terminals are designed to handle extreme weather conditions. Starlink recommends a flat high performance (FHP) terminal for users living in areas with extreme weather conditions. The earlier studies evaluated Starlink's FHP throughput for stationary and in-motion users without providing a detailed analysis of how weather affects its performance. There remains a need to investigate the impact of weather on FHP's throughput. In this paper, we address this shortcoming by analyzing the impact of weather on Starlink's performance in Oulu, Finland, a city located in Northern Europe near the Arctic Circle. Our measurements reveal that rain degrades median uplink and downlink throughput by 52.27% and 37.84%, respectively. On the contrary, there was no noticeable impact on the round-tr
Low Earth Orbit (LEO) satellite networks have attracted considerable attention for their ability to deliver global, low-latency broadband Internet services. In this paper, we present a large-scale measurement study of the Starlink network, the largest LEO satellite constellation to date. We first propose an efficient method for discovering active Starlink user routers, identifying approximately 5.98 million IPv6 addresses across 208 regions in 165 countries. Compared to general-purpose IPv6 target generation algorithms, our router-centric approach achieves near-complete coverage and, to the best of our knowledge, yields the most comprehensive known set of active IPv6 addresses for Starlink user routers. Based on the discovered user routers, we further propose an efficient method for mapping the Starlink backbone network and uncover a topology consisting of 49 Points of Presence (PoPs) interconnected by 98 links. We conduct a detailed statistical analysis of active Starlink user routers and PoPs, and further characterize the IPv6 address assignment strategy adopted by the Starlink network. Finally, we analyze the latency of Starlink user routers, propose a method to distinguish diff
The new Starlink V2 satellites incorporate improvements to the chassis brightness through dielectric mirrors, off-pointing solar arrays, and black paint on exposed components. For the general case in which the reflectivities are initially unknown, we simulate LSST operations and repeated photometry of every satellite in simulated model constellations. We derive a brightness model of the Starlink V2 satellite and study the simulated apparent brightness as a function of the satellite position relative to the observer and the sun. We find that the V2 Starlink satellites appear brightest at two distinct positions in the sky: when oriented toward the sun at low elevations where light is specularly reflected, and nearly overhead where the satellite is closest to the observer. A simulation of Starlink V2 satellites at 550 km height distributed across a series of Walker constellations with varying inclinations was analyzed to study the impact on the LSST observations. Some bright satellites will be visible in LSST observations. For every thousand V1.5 Starlink satellites imaged by LSST in the first hour of a summer night, we find 1.2 of them will appear brighter than 7 AB magnitude. By com
The development of low earth orbit (LEO) mega-constellation fundamentally threatens ground-based optical astronomical observations. To study the photometric properties of the LEO mega-constellations, we used the Xinglong 50 cm telescope to conduct a large-sample, high-precision, and multicolor target-tracking photometry of two typical LEO Mega-constellations: Starlink and OneWeb. Over a three-month observation period starting on 2022 January 1st, we collected 1,447 light curves of 404 satellites in four typical versions: Starlink v1.0, DarkSat, VisorSat, Starlink v1.5, and OneWeb. According to data statistics, Starlink v1.0 has the smallest median magnitude at clear and SDSS $gri$ band, and OneWeb is the dimmest bus. The brightness of Starlink v1.5 is slightly brighter than VisorSat. We construct a detailed photometric model with solar phase angle variations by calculating the illumination-visibility geometry based on the orbital parameters. Our data analysis shows that the solar phase angle is the significant characteristic which influencing Starlink satellites' brightness, but it is not sensitive to OneWeb satellites. VisorSat and Starlink v1.5 version, which are equipped with de
We develop signal capture and analysis techniques for precisely extracting and characterizing the frame timing of the Starlink constellation's Ku-band downlink transmissions. The aim of this work is to determine whether Starlink frame timing has sufficient short-term stability to support pseudorange-based opportunistic positioning, navigation, and timing (PNT). A second goal is to determine whether frame timing is disciplined to a common time scale such as GPS time. Our analysis reveals several timing characteristics not previously known that carry strong implications for PNT. On the favorable side, periods of ns-level jitter in frame arrival times across all satellite versions indicate that Starlink hardware is fundamentally capable of the short-term stability required to support GPS-like PNT. But there are several unfavorable characteristics that, if not addressed, will make GPS-like PNT impractical: (1) The v1.0 and v1.5 Starlink satellites exhibit once-per-second abrupt frame timing adjustments whose magnitude (as large as 100s of ns) and sign appear unpredictable. Similar discontinuities are also present in the v2.0-Mini frame timing, though smaller and irregularly spaced. (2)
The integration of satellite communication into mobile devices represents a paradigm shift in connectivity, yet the performance characteristics under motion and environmental occlusion remain poorly understood. We present the Starlink Robot, the first mobile robotic platform equipped with Starlink satellite internet, comprehensive sensor suite including upward-facing camera, LiDAR, and IMU, designed to systematically study satellite communication performance during movement. Our multi-modal dataset captures synchronized communication metrics, motion dynamics, sky visibility, and 3D environmental context across diverse scenarios including steady-state motion, variable speeds, and different occlusion conditions. This platform and dataset enable researchers to develop motion-aware communication protocols, predict connectivity disruptions, and optimize satellite communication for emerging mobile applications from smartphones to autonomous vehicles. In this work, we use LEOViz for real-time satellite tracking and data collection. The starlink robot project is available at https://github.com/StarlinkRobot.
In all networking systems, queuing is important to ensure appropriate resource utilization in the presence of bursty traffic and varying traffic demands. The Starlink access network is additionally also dynamic in terms of the capacity it can provide, and thus queuing plays an even greater role to ensure appropriate communication performance for the end-users while maintaining high resource utilization. However, for Starlink most system design details, along with the setup of the internal queuing, is private information and not publicly available. To address this we have developed a high-precision, burst-pattern controlled, traffic generation approach allowing us to precisely measure the one-way delay for Starlink. By analyzing the delay and loss in conjunction with a queue simulator we find that Starlink does not employ per-flow fair queuing or drop-tail buffers, but it does use drop-front buffer management. While drop-front reduces delay, it may also interfere with the assumptions made by loss-based congestion controls, potentially contributing to throughput degradation.
Magnitudes recorded from 2021 through 2026 for Starlink VisorSats reveal brightening of 0.6 magnitudes, while those of OneWeb show dimming of 0.4. Both trends are significant at 3 sigma. This study is based on 1.6 million magnitudes recorded by the MMT9 robotic observatory.
Low Earth Orbit (LEO) satellite ISPs promise universal Internet connectivity, yet their interaction with content delivery remains poorly understood. We present the first comprehensive measurement study decomposing Starlink's web content delivery performance decomposed across Point of Presence (PoP), DNS, and CDN layers. Through two years of measurements combining 225K Cloudflare AIM tests, M-Lab data, and active probing from 99 RIPE Atlas and controlled Starlink probes, we collect 6.1M traceroutes and 10.8M DNS queries to quantify how satellite architecture disrupts terrestrial CDN assumptions. We identify three distinct performance regimes based on infrastructure density. Regions with local content-rich PoPs achieve near-terrestrial latencies with the satellite segment dominating 80-90% of RTT. Infrastructure-sparse regions suffer cascading penalties: remote PoPs force distant resolver selection, which triggers CDN mis-localization, pushing latencies beyond 200 ms. Dense-infrastructure regions show minimal sensitivity to PoP changes. Leveraging Starlink's infrastructure expansion in early 2025 as a natural experiment, we demonstrate that relocating PoPs closer to user location red
In recent years, Low-Earth Orbit (LEO) mega-constellations have emerged as a promising network technology and have ushered in a new era for democratizing Internet access. The Starlink network from SpaceX stands out as the only consumer-facing LEO network with over 2M+ customers and more than 4000 operational satellites. In this paper, we conduct the first-of-its-kind extensive multi-faceted analysis of Starlink network performance leveraging several measurement sources. First, based on 19.2M crowdsourced M-Lab speed test measurements from 34 countries since 2021, we analyze Starlink global performance relative to terrestrial cellular networks. Second, we examine Starlink's ability to support real-time web-based latency and bandwidth-critical applications by analyzing the performance of (i) Zoom video conferencing, and (ii) Luna cloud gaming, comparing it to 5G and terrestrial fiber. Third, we orchestrate targeted measurements from Starlink-enabled RIPE Atlas probes to shed light on the last-mile Starlink access and other factors affecting its performance globally. Finally, we conduct controlled experiments from Starlink dishes in two countries and analyze the impact of globally syn
Starlink has introduced the Flat High Performance (FHP) terminal, specifically designed to support the vehicles and the vessels in motion as well as the high-demand stationary users. The research on FHP terminal throughput analysis remains limited, only a few existing studies evaluate FHP, focusing on the limited parameters and scenarios. This paper evaluates the FHP terminal's performance in Finland, Northern Europe. We examine round-trip time (RTT), uplink, and downlink throughput for both stationary and in-motion use. We measure network efficiency across six geographically diverse servers and get insights of network routing strategies. Our results show that Starlink provides high-speed, low-RTT connectivity, however, the throughput experiences fluctuations with slight degradation when in motion. Additionally, we compare Starlink and terrestrial network RTT and possible routing paths.