Conference Agenda

Broadband availability is emerging as an increasingly important dimension of communications policy as governments move beyond infrastructure deployment toward performance oversight of broadband services. Broadband is widely recognized as critical infrastructure supporting economic participation, education, healthcare, and civic engagement. Policymakers are therefore increasingly concerned not only with whether networks are deployed but also whether they remain consistently accessible to users.

Despite this growing emphasis, there is no widely accepted definition of network availability in broadband policy or measurement practice. As large-scale broadband initiatives such as the NTIA’s Broadband Equity, Access, and Deployment (BEAD) program move from infrastructure construction toward long-term oversight, defining and measuring network availability is becoming central to broadband policy.

Prior research highlights the conceptual challenges of defining broadband accessibility. Early work examining broadband as critical communications infrastructure emphasized that reliability metrics may differ depending on whether availability is measured at the infrastructure, network, or application level [1]. This observation underscores the importance of measurement frameworks that clarify which dimension of service accessibility is being observed.

Multiple approaches currently measure broadband availability, each relying on different definitions and methodologies. This paper examines whether existing broadband availability regimes influencing policymakers are converging toward a common definition or instead evaluating different layers of network operation that produce different user experience outcomes

Prior Work

Prior research recognizes that broadband availability depends on multiple components of network operation and that failures may occur at different points in communications architecture. ITU standards define service availability as the probability that a service is accessible and usable when required, distinguishing among accessibility, integrity, and retainability [2].

Internet measurement research shows that disruptions experienced by users may originate in different network segments, including home networks, access networks, routing infrastructure, transport systems, or application servers [3][4][8]. Research on broadband reliability likewise emphasizes that results vary depending on whether availability is evaluated at the infrastructure, network, or application level [1].

Although these studies highlight the layered nature of network accessibility, little research has systematically mapped contemporary broadband availability metrics onto the layered Internet architecture. This paper addresses that gap by using the layered approach as a framework to compare how different measurement systems define and measure network availability.

Methodology

This paper conducts a comparative analysis of three measurement regimes widely used by industry and policymakers to evaluate broadband reliability:

• Internet Service Providers report reliability using infrastructure uptime metrics derived from internal monitoring systems. Providers frequently advertise availability figures such as 99.99%, reflecting the operational status of infrastructure components including access networks, routing systems, and backbone facilities. These metrics measure infrastructure availability rather than users’ ability to access applications or services.
• Independent measurement platforms, including Cloudflare Radar and Outage Center [6], Ookla Downdetector [7], and OpenSignal, monitor broadband reliability using user reports, distributed telemetry, and device measurements. These systems detect disruptions at higher Internet layers, particularly transport and application layers.
• Regulatory frameworks incorporate reliability requirements into broadband performance standards. For example, networks funded through the BEAD program must meet technical performance thresholds including a 99.45% uptime requirement [5].

These approaches rely on different definitions of network availability because they observe reliability at different layers of Internet architecture. Consequently, reliability statistics derived from different measurement systems may appear inconsistent even when describing the same network conditions [9].

To address this issue, the paper introduces a conceptual framework based on the layered Internet Protocol architecture. Availability metrics are categorized according to the layer of the network stack where measurement occurs, allowing each regime to be mapped to the corresponding layer of Internet architecture where reliability is observed.

Contributions

The analysis is expected to show that existing reliability metrics measure different forms of availability at different layers of Internet architecture. Infrastructure uptime metrics primarily reflect observation of physical or media-access layers within constrained network topologies such as access networks. Independent measurement platforms capture disruptions at higher layers, including transport and application layers in end-to-end communication paths. Regulatory frameworks often combine multiple layers but lack a precise technical definition of availability.

This layered analysis may also explain why availability statistics reported by different measurement systems diverge even when evaluating the same network conditions. For example, infrastructure uptime metrics indicating 99.99% availability may coexist with application-layer disruptions affecting specific services due to congestion, routing failures, or transport-layer disruptions.

As broadband policy evolves beyond deployment toward sustained performance oversight, defining and measuring network availability consistently will become increasingly important. Clarifying how availability metrics correspond to different layers of network operation provides a foundation for measurement frameworks that better align network engineering realities with policy objectives.