For more than seven years, KASPR has continuously measured internet conditions around the globe, accumulating over seven trillion historical observations. Building on this data foundation, our predictive models and our network-aware decision technologies, we are developing a Digital World Model: a learned representation of global connectivity designed to support forecasting, the study of network dependencies, simulation and intelligent decision-making.
Our ambition is to make the digital world more observable, predictable and ultimately simulatable.
Data foundation
Understanding the digital world starts with measuring it.
Unlike models built primarily on synthetic network environments, KASPR’s research begins with a longitudinal record of observed internet conditions.
Since 2019, our measurement infrastructure has continuously tracked connectivity and network performance across countries, regions and districts, capturing both regular patterns and periods of significant disruption.
This historical depth offers an important advantage: the ability to study how network conditions have actually evolved across different geographies, providers and external circumstances.
These figures describe KASPR’s existing measurement infrastructure, which underpins the Digital World Model research programme.
Source: KASPR Datahaus (kasprdata.com), figures as published on 2026-11-04.
Beyond individual networks
Operators model the networks they control. We model the wider environment in which those networks operate.
Modern telecommunications systems are becoming increasingly intelligent. Network digital twins, foundation world models and autonomous agents are being developed to predict network behaviour and evaluate operational decisions.
These systems typically rely on detailed information about the networks they manage, including network topology, configurations, telemetry and operational actions.
Yet network performance is also influenced by conditions extending beyond individual operators: neighbouring networks, shared infrastructure, geographic conditions, changing demand and external disruptions.
KASPR brings a complementary perspective.
Our independent measurements provide a consistent, longitudinal view of connectivity across multiple providers and geographic regions. By combining these observations with predictive modelling and additional contextual information, we aim to represent aspects of the wider digital environment that are difficult to reconstruct from the perspective of a single network.
A more complete model of connectivity requires both an understanding of networks themselves and an understanding of the environment in which they operate.
Operator network models
KASPR’s modelling contribution
Capabilities
Observe. Predict. Simulate. Inform decisions.
Observe
In operationKASPR’s measurement infrastructure provides an independent view of internet conditions across geographic regions and network providers. Historical and current observations establish the empirical foundation for understanding both normal network behaviour and unusual disruptions.
Predict
In operation · extendingOur existing predictive models identify regular temporal patterns and provide information about expected network conditions. The Digital World Model research extends this work towards richer representations of network dynamics, geographic relationships and uncertainty.
Simulate
Research objectiveA central research objective is to develop models capable of representing how connectivity may evolve under different external conditions and hypothetical disruptions. Such simulations could support the study of geographic exposure, correlated network disruptions and infrastructure resilience. Their reliability must be established through comparison with observed historical events.
Inform decisions
In operation · extendingThrough KASPR Sense and our developer interfaces, we are already working to incorporate network intelligence into how people, AI agents and automated workflows plan and act. The Digital World Model will provide a foundation for progressively more sophisticated network-aware decision services, enabling applications to evaluate alternatives in light of current conditions and possible future states.
KASPR Data MCP · Digital World Model
Ask about real places, companies and networks. These Claude Code replays use existing KASPR MCP tools to investigate disruptions, compare exposure and explore a different decision.
Strategic applications
A complementary layer for the next generation of network intelligence
The development of telecom foundation world models, operational digital twins and agentic network management systems creates a growing need for reliable representations of the environments in which networks operate.
KASPR’s approach is complementary to these developments.
Our objective is not to replicate the detailed internal state of individual networks. It is to develop an independent, empirically grounded representation of global connectivity that can enrich other models and decision systems.
External network observations and historical conditions that can complement operator-specific network-state and action-response models.
Geographic and cross-network context for modelling connectivity risks, performance changes and potential external disruptions.
Observed conditions, forecasts and uncertainty information to support the reasoning and decisions of autonomous network-management systems.
A network-awareness capability that can be incorporated into applications making decisions about when, where and how connected activities should take place.
Understand changing connectivity conditions, analyse historical disruptions and explore network resilience across operators and geographies.
Assess geographic exposure to connectivity disruptions and develop more comprehensive approaches to modelling digital infrastructure risk.
Support organisations whose operations depend on connectivity across multiple locations, providers and digital services.
Inform digital infrastructure planning, disaster preparedness, connectivity monitoring and the resilience of essential public services.
Provide real-world network context to intelligent systems that must make decisions in environments where connectivity cannot be taken for granted.
Our research direction
Our approach builds on an established empirical foundation: a long-running, geographically detailed record of internet connectivity and performance, together with predictive models that are already used in KASPR’s intelligence and decision services.
The next scientific challenge is to move beyond modelling individual network indicators towards learning a coherent representation of how connectivity conditions evolve across locations, providers and time.
This involves understanding the information contained in the historical relationships between network conditions, the extent to which these relationships generalise across geographic and technological environments, and how external events influence the evolution of connectivity.
We are particularly interested in the distinction between regular variation, correlated disruptions and changes associated with identifiable external events. This distinction is important both for forecasting and for developing credible simulations of alternative network conditions.
A further objective is to link these models to decision-making. KASPR Sense already provides a practical foundation for incorporating network conditions into timing and planning decisions. Extending this towards more general decision support requires models that can characterise uncertainty and, where appropriate data are available, evaluate the consequences of alternative actions.
The research will be guided by empirical validation. We intend to benchmark new approaches against established forecasting methods, assess their performance on historical events and examine where additional model complexity produces meaningful improvements.
We also recognise the limits of externally observed data. Modelling the consequences of internal network interventions requires operational information that is not contained in KASPR’s measurements alone. Collaborations with network operators and technology partners will therefore be important for extending the scope of the model.
Our aim is to develop a scientifically grounded model of global connectivity that is useful not only for prediction, but also for understanding network dynamics and supporting decisions under uncertainty.
Building the next generation of digital-world intelligence
We are interested in research and technology partnerships that bring together complementary expertise in global internet measurement, network modelling, digital twins, simulation and autonomous systems.
