Reliability Is Not Resilience

A reliable network does not fail often. A resilient communications system continues functioning when a network does fail.

The distinction sounds semantic until a disaster exposes it. A fiber circuit with 99.99 percent availability may be excellent during ordinary operations.

If a landslide cuts the only conduit serving a valley, its historical availability is irrelevant. Two commercial cellular subscriptions may appear redundant.

If both carriers lease transport over the same damaged fiber route, the redundancy may disappear simultaneously. A building may have generator-backed Wi-Fi and cellular boosters.

If the upstream provider loses power, local backup power preserves equipment that has nothing to connect to. Resilience therefore depends on understanding failure domains.

The Communications Stack Beneath a Message

A message from one responder to another passes through more infrastructure than the handset suggests. A cellular message may move from the phone to a radio site, through fiber or microwave backhaul, into a carrier core, through authentication and routing systems, and then back toward the recipient.

A satellite connection eliminates some terrestrial access infrastructure but not necessarily the entire ground network. A land mobile radio transmission may be direct radio-to-radio, repeated through a mountaintop site or routed through a trunked system with network controllers. This is why resilience planning begins with architecture rather than brand names.

Fiber: The Invisible Backbone

Modern communications culture thinks wirelessly. The internet does not.

Fiber carries enormous volumes of traffic between cities, data centers, mobile towers, cable-headend facilities, cloud regions and international networks. Many cellular towers use fiber backhaul.

5G’s higher capacity makes backhaul even more important. Satellite gateways frequently feed traffic into terrestrial fiber networks.

Public-safety answering points increasingly use IP transport. The strength of fiber is extraordinary capacity and low latency.

Its weakness is physical geography. A fiber path can be cut by construction, flood, wildfire, bridge failure, landslide, sabotage or a damaged central facility.

The critical design issue is route diversity. Two fiber services entering the same building through the same conduit are not two independent paths.

Two providers may lease strands in the same long-haul cable. Two data centers may share the same metro transport corridor.

True resilience requires physical path knowledge. This is where communications planning begins to resemble Article 014’s mineral-provenance problem: the label on the service is less important than the chain beneath it.

Cellular and the FirstNet Experiment

FirstNet is one of the largest attempts ever made to combine commercial cellular economics with dedicated public-safety capabilities. The network emerged from a lesson of September 11: emergency responders need interoperable communications and should not be treated like ordinary consumer traffic when networks are congested.

FirstNet uses AT&T’s nationwide network while adding dedicated Band 14 spectrum, public-safety priority and preemption, a dedicated network core and a fleet of deployable assets. The architecture has expanded substantially.

NTIA announced in March 2026 a revised FirstNet/AT&T framework intended to unlock roughly $2 billion in additional public-safety value, including acceleration of a dedicated 5G core. FirstNet has separately outlined a multibillion-dollar transition toward 5G, mission-critical services and expanded coverage.

Its current coverage program adds more than 135 purpose-built sites on top of 1,000 sites added during 2024 and 2025. FirstNet says it has also deployed more than 14,000 miniature indoor cells.

The most strategically interesting addition is future satellite-to-device service. FirstNet says its dedicated core is being upgraded to support public-safety-specific satellite-to-device capabilities including priority and preemption on Band 14 and mission-critical push-to-talk. That would move public-safety broadband toward a hybrid architecture:

This is exactly the direction a resilient communications system should move. The question is whether the layers are sufficiently independent in real failures.

The 2026 FirstNet Reality Check

A network built for emergencies needs stronger evidence than a commercial coverage map. The Department of Commerce Inspector General supplied an uncomfortable counterweight in April 2026.

Its audit concluded that FirstNet Authority did not ensure the Nationwide Public Safety Broadband Network met service-availability requirements. The OIG found that the Authority’s measurement approach covered only a fraction of cell sites and of the network’s roughly three-million-square-mile coverage footprint. It also said FirstNet did not adequately verify contractor-provided information or service-availability requirements for Pacific territories.

The audit produced four recommendations addressing clearer performance standards, availability requirements, surveillance methods, auditing and remediation. This matters for reasons beyond FirstNet.

Communications resilience is difficult to measure. A network can show excellent national availability while failing at precisely the location where a disaster occurs.

A cell site’s radio can be healthy while backhaul is unavailable. A network can be technically reachable but too congested to support useful operations.

A carrier can restore consumer service while an emergency organization still lacks the applications, credentials or local coverage required for its mission. The correct metric is therefore not only uptime.

It is mission availability. Could the intended user communicate, with the required service, at the required location, during the actual failure?

Why Land Mobile Radio Refuses to Disappear

Broadband has transformed public safety. Maps, body-camera video, databases, telemetry, location services, photographs and incident-management applications require more than traditional voice radio can provide.

Yet land mobile radio remains difficult to replace. Dedicated public-safety LMR systems are optimized around immediate group voice.

Users push a button and talk.

Networks can be designed for specific geography. Systems may support direct device-to-device operation when infrastructure is unavailable.

Repeaters can be hardened and strategically located. Public-safety users understand the operating model.

CISA continues to treat LMR as a core part of the emergency-communications ecosystem while describing a long-term convergence with wireless broadband rather than an immediate replacement. The Maui wildfire provides a harsh real-world example.

Associated Press reporting found that cellular infrastructure and fiber were heavily damaged in Lahaina, while traditional radio remained usable for fire crews through a repeater located away from the affected area. That does not mean LMR is invulnerable.

Repeaters need power.

Trunked systems have control infrastructure. Radio sites can burn or flood.

Encryption and interoperability require planning. But a communications architecture that preserves LMR alongside broadband creates a genuinely different failure path.

Deployable Networks

When fixed infrastructure fails, mobility becomes resilience. FirstNet’s deployable fleet is built around that principle.

Satellite Cell on Light Trucks function as mobile cellular sites with satellite backhaul. Compact and mini rapid deployables can be moved into remote areas.

Tethered aerial cells can raise antennas above terrain. Low-Earth-orbit emergency communication portables can restore backhaul when fiber or microwave paths fail.

The Nashville bombing demonstrated both the dependency problem and the value of deployables. The explosion damaged an AT&T central office. Equipment initially continued on batteries, but flooding disabled generators and batteries eventually exhausted. FirstNet reported that satellite-linked deployable cells restored local public-safety connectivity and that seven SatCOLTs were operating within twenty-four hours.

The lesson is not that deployables solve everything. They introduce new constraints:

travel time;

fuel;

trained operators;

satellite visibility;

spectrum;

local site selection;

authentication;

capacity.

A deployable asset sitting three states away is resilience only if it arrives before the operational need is over. This is why distributed caches of smaller deployables can sometimes matter more than a few spectacular large systems.

Satellite Moves Into the Phone

Article 008 examined satellite internet as an emerging second global communications layer. Article 015 asks a narrower question:

what happens when satellite becomes one path inside the ordinary mobile network? 3GPP Release 17 created the first normative 5G requirements for non-terrestrial networks. The standards explicitly address service continuity between terrestrial and satellite access networks, roaming and satellite-supported handheld devices.

That standards work is now appearing in commercial systems. Starlink Mobile is expanding direct-to-cell service and has acquired additional spectrum as SpaceX signals broader mobile ambitions.

AST SpaceMobile is deploying large LEO satellites intended to connect directly to standard smartphones. Amazon Leo has proposed a separate direct-to-device constellation.

Traditional mobile-satellite operators and systems such as Globalstar, Iridium, Skylo-linked networks and others are also moving toward deeper integration with consumer devices. The ITU describes D2D as an emerging layer of global telecommunications infrastructure with particular value in dead zones, disasters and remote areas.

The resilience value is enormous.

A terrestrial tower can be destroyed while the satellite remains overhead. But satellite resilience has boundaries.

Buildings block sky visibility.

Dense urban canyons degrade links.

Handset transmit power is limited.

Spectrum is finite.

One satellite beam may cover a huge population. The system may still depend on terrestrial gateways and carrier cores. Satellite is therefore a second path—not a magical replacement for terrestrial communications.

Coverage Is Not Capacity

This distinction may be the most important technical lesson in direct-to-device communications. A user who can send a text from a wilderness area has coverage.

A stadium with 60,000 users requires capacity. A disaster zone can have both problems simultaneously.

Satellite beams cover far larger geographic areas than terrestrial cells. That gives them extraordinary reach but forces many users to share limited radio resources.

Terrestrial cellular networks solve capacity by densifying: more towers, more sectors, more spectrum, more fiber. A satellite cannot simply add a street-corner small cell.

This makes D2D especially valuable for: messaging; SOS; basic voice; low-rate data; continuity when nothing else exists. It should not automatically be described as a replacement for urban terrestrial broadband.

Independent measurement research published in late 2025 reinforces the value of heterogeneity. Researchers measuring Starlink, OneWeb and terrestrial 5G found that combining independent networks substantially improved outage performance in their test environments. Their specific figures apply to the studied setup rather than every deployment, but the direction is important: multi-connectivity can be more resilient than optimizing one network alone.

The Power Dependency

Communications systems fail electrically before they fail philosophically. Every layer requires energy.

A smartphone requires a charged battery. A radio site requires DC power.

A fiber terminal requires electricity. A router requires electricity.

A satellite terminal requires electricity. A repeater requires electricity.

A network core requires data-center power. The Nashville outage showed the cascade vividly: batteries preserved service after utility power failed, but flooding disabled generators, batteries eventually depleted and communications failed.

Communications resilience therefore requires an energy plan: batteries; generators; fuel contracts; solar where appropriate; portable power; load shedding; site hardening; and the ability to refuel during regional disruption. This is why Articles 006, 011, 012 and 013 connect directly to communications. A communications system without resilient power is merely a network waiting to become unavailable.

The Core-Network Dependency

Modern communications systems are increasingly software-defined. That creates flexibility.

It also creates central dependencies. A mobile network relies on authentication, subscriber databases, routing, policy, billing or entitlement systems, security infrastructure and increasingly cloud-native network functions.

A local tower can be physically intact while a distant core failure makes service unusable. FirstNet’s dedicated core exists partly to separate public-safety traffic and provide priority, preemption and mission-specific functions.

The planned standalone 5G core expands that model. Resilience therefore has to be assessed at two scales:

local survivability and systemic survivability. A local disaster can destroy towers and fiber.

A software or cyber incident can affect a geographically distributed core. Article 010’s cybersecurity lesson returns here: a physically redundant communications system can still fail through one shared identity or software control plane.

Mesh and Local Autonomy

Mesh networking is attractive because it appears to eliminate infrastructure. Devices relay information through one another.

If one path disappears, another can form. That can be useful for local teams, sensors, temporary incidents, military systems and specialized community networks.

But “mesh” is not synonymous with unlimited resilience. A mesh has range limits.

It needs enough participating nodes. It may have low throughput.

Routing overhead increases.

Encryption and identity still matter. And a local mesh that ultimately needs internet access still needs some gateway out.

The most useful role is often local autonomy: maintain communications inside the incident area even while long-haul connectivity is degraded. Then bridge outward through whichever backhaul remains:

fiber;

cellular;

microwave;

satellite.

Priority Is Not Extra Capacity

During disasters, communications networks can remain physically intact and still become unusable through congestion. Everyone calls at once.

Everyone uploads video.

News crews arrive.

Families attempt to reach one another. CISA’s priority telecommunications services and FirstNet’s priority/preemption functions exist for this problem.

GETS gives eligible users prioritized processing through landline and related telephone networks. Wireless Priority Service provides priority on participating cellular networks. FirstNet gives public-safety traffic priority and, in defined circumstances, preemption over lower-priority traffic.

These mechanisms are valuable.

They do not manufacture spectrum or backhaul. If a site has no power, priority cannot restore it.

If a satellite beam has limited throughput, priority can allocate scarce capacity but not create unlimited capacity. If every path shares one severed fiber route, preemption cannot repair the fiber.

Priority solves congestion.

Redundancy solves path failure. They are different resilience controls.

PACE and Failure Independence

CISA uses the PACE model—Primary, Alternate, Contingency, Emergency—as a framework for continuity planning. The idea is deceptively simple.

What is the normal method?

What replaces it?

What happens after that fails?

What is the last-resort capability? The mistake is filling each box with technologies rather than failure domains.

A weak PACE plan might look like:

Primary: Carrier A cellular.

Alternate: Carrier B cellular.

Contingency: FirstNet smartphone.

Emergency: cellular hotspot.

That is four labels describing essentially one architecture. A stronger model might separate mechanisms:

This is the principle of failure-independent PACE. Each successive layer should remove at least one major dependency that could destroy the previous layer.

The SURVXCOM Communications Resilience Test

Article 008 introduced a communications-resilience framework for satellite and network systems. Article 015 expands it into a full operational test.

What to Watch Next

The Network That Survives Failure

The future of resilient communications is not a bigger cell tower. It is not Starlink.

It is not FirstNet.

It is not land mobile radio.

It is not fiber.

It is not mesh.

It is not a satellite phone.

It is the relationship among all of them. The communications system that survives is usually the system that assumes something will break.

The tower may lose power.

The fiber may be cut.

The carrier core may fail.

The satellite may be blocked by terrain. The radio repeater may be damaged.

The internet may be congested.

The operator may forget the fallback procedure. Resilience begins when planners stop asking:

Which network is the best?

and start asking:

What still works after this network is gone? That question changes procurement.

It changes training.

It changes backup power.

It changes where fiber enters a building. It changes whether agencies preserve LMR.

It changes whether satellite is stored as emergency equipment or integrated continuously into the network. It changes whether a portable cell is pre-positioned or requested after disaster.

It changes whether a communications exercise tests normal operations or deliberately kills the primary network. Most importantly, it exposes fake redundancy.

Two carriers sharing one conduit are not two paths. Two applications sharing one cloud are not two systems.

Three radios using one repeater are not three networks. Satellite backhaul through one powerless gateway is not independence.

The communications architecture of the next decade will become increasingly sophisticated as terrestrial broadband, 5G cores, non-terrestrial networks, direct-to-device service, deployables and software-defined systems converge. That convergence should increase resilience.

But only if architects preserve difference. The goal is not one universal network. The goal is a communications stack in which no single failure gets the final word.