IAB Workshop on IP Address Geolocation (ip-geo) (ipgeows)

Daniel Schatte, Charter Communications

From an ISP perspective, IP-geolocation serves important operational purposes, but it also raises significant privacy concerns that must be addressed. ISPs sit at the intersection of technical necessity and customer trust: while we facilitate access and provide the infrastructure, we also bear responsibility for protecting the privacy of millions of subscribers.

The comments focus on the dual priorities of supporting legitimate operational use cases while ensuring that geolocation practices do not compromise end-user privacy.

Mudassar Tariq

IP address geolocation shapes how the Internet adapts to place. It answers “where is this IP?” so apps can localize content, steer CDNs, flag fraud risk, enforce com- pliance, and measure performance. Providers use it for everything from routing and coarse country blocks to fine-grained controls like fraud prevention, ad targeting, and regulatory gating.

Methods range from latency measurements to triangulate approximate location on multiple signals, but lag when ISPs reassign or renumber prefixes, causing misrouting, bad geoblocks, and poor targeting that lead to user complaints and service issues. To close the gap, operators publish geofeeds, a simple standard mapping prefix to locations. Large platforms consume these feeds, for example Google, and many ISPs now publish them, which update location data faster and more accurately than static databases.

Deployments still break in predictable ways, including over-broad claims, conflicts on leased space, dead discovery links, duplicate CIDRs, bad admin codes, and uneven refresh and attestation. Demand for precise, current data fuels a busy geolocation market. This paper explains why today’s ecosystem fails in predictable ways and proposes implementable practices to convert IP-geo from a “best-effort guess” into a trustworthy coarse signal.

Richard Barnes

There are an increasing number of legal requests for operators of Internet applications to change their behavior. For example:
● Several jurisdictions have imposed age-based restrictions on content availability. For example, a law in the US state of Mississippi requires operators of websites that allow user-generated content to implement age verification and block access for users under the age of 18. In response, the Bluesky social network disabled service in Mississippi.
● DNS resolver operators regularly receive demands to block certain domains. In 2024, a court in France ordered major DNS resolver operators to block certain sports streaming sites. In response, Google blocked the sites in question, and OpenDNS disabled service in France entirely.
● Recent revisions to the German telecommunications law require operators of telecommunications service to provide lawful intercept capabilities. The definition of telecommunications services explicitly includes number-independent interpersonal communications services, i.e., over-the-top services.

Zoltan Szamonek

IP address geolocation (ip-geo) is currently the default for determining a device's real-world location, used for everything from content optimization to legal compliance. However, IP addresses were not designed for this purpose, leading to inherent inaccuracies and significant privacy concerns. This paper argues that long term continued reliance on ip-geo is unsustainable, especially with the rise of privacy-preserving network architectures. We advocate for standardizing a non-IP-based, user-controlled location signal that provides both the utility of coarse-grained location and the necessary trust for compliance, without compromising user privacy.

Yoshiki Ishida: JPIX Co., Ltd./IPoE Counci , Hiroki Kawabata: JPNIC , Kazuma Kazama: Geolocation Technorogy, Inc. , Akihiro Mizuochi: Broad Band Tower, Inc, Nobuhiro Takamizawa: J-Stream Inc.

In Japan, Internet Service Providers (ISPs) offer internet connectivity over NTT’s nationwide NGN (Next Generation Network). This infrastructure allows IPv6 addresses to be partially linked to geographic information. Furthermore, a database that maps IP addresses to geographic locations is maintained by representatives of the ISPs.
Taking this situation into account, this report outlines the current state and challenges of GeoIP in Japan, as well as its potential future applications.

Thomas (Tommy) Croghan (Lost Creek Technology, LLC The Brothers WISP Podcast)

As a consultant for numerous small to medium-sized Internet Service Providers (ISPs), I am writing to highlight the significant operational challenges we face regarding the current state of IP address geolocation. These issues not only impact our ability to deliver a consistent and reliable user experience but also consume valuable resources in a constant, often frustrating, battle against outdated and inaccurate data. The core problems are the lack of a standardized, timely feedback mechanism for corrections, the persistent misattribution of our networks, and the prolonged propagation delays for geolocation updates.

Siddharth “Sid” Mathur

Fastah Inc. provides IP geolocation API services to developers worldwide. We ingest and process geolocation feeds from IP owners, and we also help developers build their use cases with personalized support and reliable, high-performance API operations. Fastah is a strong adopter of RFC 8805 and RFC 9632, and is rated highly on GeolocateMuch.com.

Robert Kisteleki, RIPE NCC

Most contemporary IP geolocation services and approaches focus on end-user (“eyeball”) geolocation - mainly because that’s where the most understood business need is; payment process verification, geofencing, content localisation heavily depend on such services.
However, there’s also a visibly growing need to geolocate infrastructure components. As of yet this is a much less supported case.

Ricky Hosfelt

Fastly submits this position paper in response to the announcement by the program committee of the IETF’s interest in convening the IAB Workshop on IP Address Geolocation in December 2025
.
As both a Content Delivery Network (CDN) and Edge Compute Platform, Fastly has a deep interest in IP address geolocation from the dichotomy of being both a consumer and provider. Fastly currently utilizes IP address geolocation to enrich customer connections through its CDN, and provides an API endpoint for querying geolocation information against a data set from both its platforms. Fastly customers rely on the accuracy of geolocation services and the consumption of RFC 88052 geofeeds to represent the locations of proxied users.

Om Prakash Sharma

As an application developer and AI/ML engineer, I frequently integrate IP address–based services into real-world applications, ranging from educational platforms and content delivery systems to cloud-native applications deployed across multi-cloud environments. My experience spans both user-facing applications and infrastructure-level orchestration, which has given me insight into the strengths and shortcomings of IP geolocation today.
I am particularly interested in contributing to this workshop from the perspective of a developer building applications for diverse and bandwidth-constrained regions such as Nepal, where user privacy, localization accuracy, and accessibility are often at odds with each other.

Oliver Gasser, William Leung, Maxime Mouchet (IPinfo)

IP address geolocation has many different applications in today’s Internet, ranging from website language selection, ad targeting, fraud detection, to content geofencing. Recently, geofeeds have been proposed as a way for network operators to publish geolocation information about their own IP prefixes. Geofeeds are CSV files which contain geolocation information about IP prefixes, containing fields for country, region, and city information. Geofeeds can be shared through dedicated entries in the WHOIS database or privately with geolocation service providers. Generally, geofeeds have had a positive impact on geolocation sharing from network operators with geolocation providers by providing a standard interface for exchanging data. There remain, however, a number of challenges when working with geofeed data today, which we discuss in this paper.

Nathan Owens, Starlink Networking

Navigating challenges in IP Geolocation (GeoIP) is critical in operating Starlink, a global internet service provider that leverages a constellation of over 8,500 satellites in low earth orbit (LEO). Starlink provides service to millions of users across 150+ countries and territories, with the capability to support users in every remote region in the world, including aviation and maritime use cases. This position paper discusses the use of and problems with GeoIP, along with recommendations for improvements, drawing from the Starlink network experience over the past 5 years.

Nalini Elkins, Outside the Stacks, Inc.
Mark Nguyen, Outside the Stacks, Inc.
Bill Jouris, Outside the Stacks, Inc.

IP geolocation data is a critical tool in the arsenal of governments, financial institutions, and security platforms defending against cyber actors and transnational fraud which may be tracked to specific geographic regions.  However, the current mechanisms for identifying the geographic origin of IP traffic suffer from key gaps: poor accuracy, manipulation through anonymization techniques (e.g., VPNs, proxies), dynamic IP reassignment, and insufficient integration with threat intelligence. This paper briefly highlights how adversaries from North Korea, Iran, Russia, and China have exploited these weaknesses, including tactics like remote work deception, infrastructure hijacking, and cloud service abuse. We recommend redesigning geolocation strategies to include contextual risk modeling, endpoint verification, and multi-signal attribution.

Md. Kamruzzaman Khan

Current IP geolocation systems operate as pervasive surveillance infrastructure, tracking users without consent and violating fundamental privacy principles. This position paper proposes a paradigm shift from surveillance-based to consent-based geolocation, addressing key IAB workshop topics including trust and privacy issues, gaps in current approaches, and alternative solutions. We present a Privacy-First Geolocation Protocol (PFGP) that enables user-controlled location sharing while meeting legitimate business requirements for CDNs, streaming platforms, and other applications.

Marwan Fayed

IP geolocation has its origins in the 1990s, at a time when IP addresses were long-lasting and often statically located in networks, two attributes that are no longer reliably true of IP addresses today. It also predates large-scale NAT devices, ubiquitous use of VPNs and other IP-sharing mechanisms, alongside mobile devices and their networks. Underlying geolocation mechanisms have been refined and improved, but are otherwise unchanged. In 2025 and beyond, it behooves us to ask what question IP geolocation truly answers, then identify the set of actual questions that need answers — and the extent to which they are answered using IP geolocation information.

Lee Howard

IP geolocation starts from the flawed assumption that an IP address is associated with an interface of a device that has a specific location in space. Even such trivial and common technologies as VPNs and containerization create a mismatch between the expected location and the intent of the geolocation user. Worse, IPv4 transfer and leasing markets make IPv4 addresses even more mobile than traditionally expected. Few of the major use cases support making a concerted effort to improve the system, and there may be dangerous unintended consequences.

Kevin Vermeulen (LIX, CNRS, Ecole Polytechnique)

IP geolocation and academic research is a long-told story. And yet, after decades of research, it is still an open problem, and there exists no public, accurate, large-scale, IP geoloca- tion dataset coming from the academic world. However, IP geolocation is one of the most used metadata in the com- munity, and researchers would love to have access to such a dataset. As both recent users of IP geolocation data and contributors to design of new IP geolocation techniques, we share in this paper our experience on how the research com- munity apprehends geolocation, and open a discussion on research on IP geolocation and its usage and how we could improve it.

Katherine Izhikevich (UC San Diego), Sumanth Rao (UC San Diego),
Ben Du (UC San Diego), Alisha Ukani (UC San Diego), Manda Tran (UC Los Angeles), Liz Izhikevich (UC Los Angeles)

To geolocate an IP address, current methodologies often rely upon a global, human-reported set of vantage points (i.e., probes) to measure from and verify accuracy against. In this work, we analyze how often and where human-reported geolocations are incorrect.

Joe Abley, Cloudflare

Cloudflare supports the development of an open, interoperable, secure and complete framework
for the management of geolocation data and stands ready to contribute and participate.

Jianping Pan and Jinwei Zhao

Low-earth-orbit (LEO) satellite networks, exemplified by SpaceX’s Starlink, Eutelsat’s OneWeb, Amazon’s Project Kuiper and Telesat’s Lightspeed and followed by others, have revolutionized the Internet access, particularly in rural areas, in the air and on the sea, where traditionally Internet access technologies are unavailable, too costly, very limited or simply impossible. Besides lower latency and higher capacity on the ground, LEO networks also introduce new challenges such as frequent satellite handover. For example, Starlink hands over every 15 seconds, fixed at the 12th, 27th, 42nd and 57th seconds off each minute. With a different satellite and likely ground station, the latency between user terminal (i.e., dish) and user’s home point-of-presence (PoP) changes greatly, which confuses TCP congestion control algorithms (i.e., a sudden latency increase leads to premature timeout and decrease results in packet reordering and duplicate acknowledgment). TCP and QUIC adjustment and bitrate adaptation are used to improve protocol and application performance over LEO networks.

Jeff Brown

Content Delivery Networks (CDNs) have long relied on IP address geolocation (geo-IP) for traffic routing and content optimization decisions. However, the implementation of geo-IP usage varies significantly across providers, with substan- tial implications for user privacy and network performance. While Apple’s CDN infrastructure is an example of geo-IP usage that addresses current operational
needs without compromising sensitive data, CDNs are evolving beyond exclu- sively using geographic inference. For example, there are emerging standards like Service Binding (SVCB) record-based routing1 that can eliminate server-side location inference entirely. This would allow CDNs to shift decision-making from server-side geo-IP lookups to client-side service selection. Other techniques such as network performance-based routing2, differential privacy applications3, and user-controlled location disclosure4 would allow CDNs to improve performance and reliability while providing guarantees of user privacy protection.

James Clark

GeoComply is a Vancouver-based cybersecurity and fraud prevention company that specializes in geolocation technology. Initially, the company's patented and proprietary solutions were developed for the highly regulated U.S. online gaming and sports betting market to ensure compliance with jurisdictional laws. GeoComply has since expanded its services to other sectors, including financial services, cryptocurrency, and media and entertainment, with the goal of creating a safer internet by instilling confidence in online interactions and protecting businesses from fraud and non-compliance penalties. GeoComply's expertise in geolocation goes beyond simple IP address checks, which can be easily manipulated. Its technology combines multiple data sources, including Wi-Fi, GPS, cellular data, and IP addresses, to pinpoint a user's true location with high accuracy. The company also employs advanced anti-fraud techniques to detect and combat location spoofing methods including VPNs, proxies, and advanced techniques such as fake GPS apps, emulators, and device tampering.

Geoff Huston

At APNIC Labs we have been investigating if the collection of data that we have assembled as part of the measurement work can be used to track the ISP market share within each national economy. We are interested in trying to measure the effective level of inter-ISP competition within each national economy. The base of this derived competition measurement is a notional count of end users that are served by each ISP that operates in a national economy.

Augustin Laouar (École Normale Supérieur de Lyon), Paul Schmitt (California Polytechnic State University), Loïc Desgeorges (Université Claude Bernard Lyon 1), Francesco Bronzino (École Normale Supérieur de Lyon)

Location underpins critical Internet services, yet our primary mechanism for Internet localization, IP-based geolocation, fails to meet the needs of all stakeholders. User location is conflated with network location, leading to a fundamental mismatch between the goals of content providers, infrastructure operators, and regulators. As users increasingly adopt privacy-preserving technologies that obscure their network identity, this mismatch becomes more pronounced, making localization even more challenging. We argue that the problem cannot be solved by simply improving the accuracy of incumbent mechanisms that are inappropriately applied today to solve multiple, unrelated problems. Instead, we require a new approach for localization on the Internet.

Erik Nygren with contributions from Rizwan Dhanidina

IP:Geo location databases and systems are an important part of the Internet ecosystem. They have their inherent limitations, but they are not immediately replaceable. Akamai has our own IP:Geo location product (“EdgeScape”) that we both sell directly, use internally, and have incorporated into a number of our other products and services. Akamai also publishes multiple public IP:Geo feeds in RFC 8805 format for address space that we control.

Paul Gao, Enock Lee, Yaozhong Zhang
Beijing Tiantexin Tech. Co., Ltd.(IPIP.NET)

1. How to use IP-geo data?
2. What problems exist with the industry's current production methods?
3. What problems exist in the IP geolocation data currently used in the industry?
4. Is it necessary to establish a unified format and standards for publishing and updating IP geolocation data?
5. How is IP geolocation different from other geolocation methods?
6. Are there alternative methods besides IP-based positioning?
7. In an IP database, besides geographic information, is it valuable to convey other information?
8. How to define coverage?
9. How to define accuracy?
10. What kind of commercial licensing model should an IP database adopt?

Divyank Katira, Internet Research Lab; Gurshabad Grover, Internet Research Lab\

The primary function of the IP address is to identify and route to entities that are reachable through the internet. As RFC 791, an early specification, states: “The internet protocol provides for transmitting [...] datagrams from sources to destinations, where sources and destinations are hosts identified by fixed length addresses.”

Since this early conception, IP addresses have come to be used in a number of other ways – primarily to glean information about end-users. This includes profiling internet users for behavioral advertising, abuse prevention and law enforcement; building IP reputation systems for spam and DDoS prevention; and geolocating users for localization, optimised service delivery and to comply with local laws.2
While not being designed for this purpose, the use of the IP address as a stand-in user identifier and for deriving information about a user has served as a quick-and-easy way for many business, security and legal use cases. These uses have found implicit and explicit support in various standards-setting processes.

We argue here that IP geolocation is an instance of sensitive and private data being generally abused to many ends, including for privacy violations and censorship. IP geolocation happens without user consent, and should be phased out in favour of privacy-respecting alternatives.

Debayan Mukherjee

In this digital era, IP geolocation has become a very critical component. In modern network operations and content delivery and security enforcement, and regulatory compliance. However, current approaches suffer from fundamental limitations that hinder accuracy, standardization and practical deployment. This paper identifies the gaps and the problems of existing IP Geolocation approaches.

T. Pauly (Apple Inc.), D. Schinazi (Google LLC), C. McMullin (Google LLC), D. Mitchell (Google LLC)

Techniques that improve user privacy by hiding original client IP addresses, such as VPNs and proxies, have faced challenges with server that rely on IP addresses to determine client location. Maintaining a geographically relevant user experience requires large pools of IP addresses, which can be costly. Additionally, users often receive inaccurate geolocation results because servers rely on geo-IP feeds that can be outdated. To address these challenges, we can allow HTTP clients to actively send their network geolocation to an HTTP server via an HTTP header field. This approach will not only enhance geolocation accuracy and reduce IP costs, but it also gives clients more transparency regarding their perceived geolocation. This is also particularly useful in the case of HTTP intermediaries that hide client IP addresses, such as Oblivious HTTP (OHTTP) relays.

ALAGAPPAN RAMANATHAN, University of California Irvine, USA;  SANGEETHA ABDU JYOTHI, University of California Irvine, USA

Traceroute and IP geolocation are essential tools for Internet infrastructure analysis, supporting applications from network topology mapping to censorship detection. However, when these tools are used together in practice, their individual limitations compound to create systematic inaccuracies that significantly impact measurement reliability.

To understand how these limitations compound, we examined each tool individually. IP geolocation aims to map IP addresses to physical locations, but commercial databases often disagree substantially—past studies show accuracy as low as 33% at the country level for some databases. Simultaneously, traceroute measurements face challenges from network infrastructure elements like MPLS tunnels and interface address variability that can obscure true network paths. When researchers combine these imperfect tools to analyze network topology and performance, the resulting errors propagate throughout their analyses.

Our preliminary analysis of a day-long traceroute measurement from RIPE Atlas demonstrates that approx- imately 5.4% of IP addresses in this dataset exhibit what we term "anomalous geolocations"—locations that are inconsistent with network measurement data. These anomalies affect at least one hop in 55% of traceroutes and 20% of network links, creating widespread contamination of network analysis results.