Analyzing OneWeb’s LEO Satellite Network: Connectivity, Performance, and Challenges

The rise of Low Earth Orbit (LEO) satellite networks is revolutionizing global connectivity, providing low-latency, high-throughput Internet access to remote and underserved areas worldwide. OneWeb, one of the leading names in the LEO space, operates the second-largest commercial satellite network, focusing on enterprise and government markets. Unlike its primary competitor, Starlink, OneWeb’s real-world network performance has remained relatively under-studied—until now. Recent research sheds light on its infrastructure, performance, and unique challenges in delivering seamless connectivity across the globe.

How OneWeb Stands Apart from Starlink

As of August 2025, Starlink operates over 8,000 satellites, mainly positioned at an altitude of 550 km. These satellites use a diverse range of inclinations to ensure global coverage, including high-latitude areas. In contrast, OneWeb operates a comparatively modest constellation of 650 satellites at an altitude of 1,200 km. These are distributed across 12 near-polar orbital planes with an inclination of 87.9°. This unique configuration results in denser satellite coverage in polar regions but sparser coverage near the equator.

Another key difference between the two lies in their use of inter-satellite links (ISLs). Unlike Starlink, which utilizes ISLs for seamless data transfer, OneWeb relies heavily on terrestrial ground infrastructure. OneWeb’s network is supported by 29 Points-of-Presence (PoPs) and 40 Satellite Network Portals (SNPs) scattered worldwide. However, this lack of ISLs poses significant challenges during satellite handovers, particularly in maintaining low-latency connections as packets traverse long stretches of terrestrial fiber networks.

Performance Insights: Latency and Throughput Analysis

A notable finding of the study is how OneWeb manages latency during satellite handovers. Researchers observed that transitions between satellites can introduce significant variations in round-trip time (RTT). For instance, network packets initially routed through the Santa Paula SNP could experience higher latency as they traverse across the continent. A subsequent satellite reconfiguration, switching to the Southbury SNP, notably reduced latency by almost half. These handover dynamics create predictable bimodal latency patterns, opening the door for machine learning algorithms to optimize performance further.

When measuring throughput performance, OneWeb consistently met its guaranteed levels but exhibited variance depending on the transport layer protocol and congestion control algorithms employed. For example, while a UDP stream maintained steady performance, TCP throughput suffered fluctuations based on the algorithm used, with the BBR algorithm emerging as the most effective. Such findings provide valuable insights into the intricate balance between network design and protocol optimization in satellite-based Internet services.

The Challenges and Opportunities of OneWeb’s Approach

The absence of ISLs and reliance on terrestrial PoPs remain key operational challenges for OneWeb. Unlike satellite-based routing, this design increases latency and complicates transitions during satellite handovers, particularly in sparsely covered regions. Despite these hurdles, OneWeb’s infrastructure offers promising opportunities for refining latency models and further advancing transport layer optimizations. The predictable nature of latency patterns, arising from the Earth’s rotation and satellite orbits, could be leveraged to enhance real-time services and adaptive applications, such as video streaming.

As LEO satellite networks continue to evolve, the insights from studies like this one will prove invaluable in shaping the future of global internet connectivity. By addressing existing technical limitations, OneWeb has the potential to further enhance its services, ensuring reliable and efficient Internet access for even the most remote regions of the world.