the national warning system

Previously on Deep Space Nine, we discussed the extensive and variable products that AT&T and telephone operating companies sold as private lines. One of the interesting properties of private line systems is that they can be ordered as four-wire. Internally, the telephone network handles calls as four-wire with separate talk and listen pairs (or at least, it did before digitization). For cost reasons, though, service to individual customers is virtually always two-wire, with talk and listen combined onto a single pair via hybrid transformers. Four-wire private lines are just about the only exception.

Why? Well, one of the major advantages of four-wire service to the telephone instrument is that it avoids the echo and sidetone that normally occur within the hybrid transformers. On a call between two telephones, this effect is acceptable and even desirable. In conference systems, though, with many phones attached, echo accumulates until the line is almost unusable. Prior to the introduction of DSP technology to "clean up" the audio, multiparty conferences were a lot more limited than we take for granted today... except for the four-wire private line systems specifically built for large conference calls. The most notable of these is the National Warning System, or NAWAS, operated by AT&T for the Federal Emergency Management Agency (FEMA).

FEMA has an interesting history. It is most directly a product of the Department of Housing and Urban Development, where was originally established in 1973 with the responsibility for coordinating reconstruction after natural disasters. Over time, a series of federal reorganizations expanded FEMA and added additional roles. Most notably, in 1979 President Jimmy Carter instituted a major reorganization of federal emergency agencies that dissolved the Office of Civil Defense, made FEMA an independent agency, and placed all civil defense responsibilities within FEMA. As a result, part of FEMA is a direct descendent of the civil defense efforts at the peak of the Cold War. FEMA operates the government relocation bunker at Mt. Weather, for example, and by the same coin is responsible for the dissemination of attack warnings to the contiguous United States.

The origins of NAWAS can be traced back to the Civil Defense Warning System (CDWS), often known as the "Bell and Lights System," which was introduced in the 1960s and is itself an extension of some earlier precedents. The various iterations and renaming of NAWAS make it a little bit difficult to trace its history exactly. Wikipedia says that NAWAS was formed in 1978, a reasonable claim given that FEMA organized around that same time. But NAWAS cannot have been new in '78: AT&T published a BSP covering the "OCD NAWAS," prior to FEMA's existence, a full decade earlier in 1968. Indeed, NAWAS and the CDWS or "Bell and Lights" must have operated in parallel, as both had BSPs issued that same year.

That's not actually that surprising: one of the reasons that the nation's emergency communications networks kept being replaced is because the requirements kept changing. CDWS was designed primarily as an automated or machine-to-machine system, capable of activating air raid sirens and sounding local alarms automatically. During the height of the Cold War this was a good fit for the intent. The emergency scenario was nuclear attack, and a warning would need to be disseminated as rapidly as possible for optimum lifesaving effect. The Office of Civil Defense once targeted a 30-second timeline from declaration of an alert to the American public becoming aware.

Even as CDWS was put into service, though, OCD was aware that a more extensive communications capability would be required to distribute information on attack outcomes, evacuation and recovery efforts, and to enable continuity of government even in a possible scenario of devolution of control to local emergency response authorities. That need would be served by NAWAS: not an alarm system, but a voice communications system, ready for two-way use on state, regional, and national scales.

Over time, FEMA has shifted its emergency alert programs away from nuclear conflict and towards the "All-Hazards" model, in which the scope of the systems is interpreted broadly and the primary use tends to be natural disaster and weather alerts. The "All-Hazards" concept came about mostly because of a sense that FEMA and the National Weather Service were pointlessly duplicating capabilities; so because of All-Hazards thinking, FEMA NAWAS distributes critical weather alerts and NWS's weather radio network distributes FEMA alerts. NAWAS has thus changed its identity from a wartime system to a more general emergency management system.

Let's see how NAWAS actually works. NAWAS is, at its core, a network of interconnected four-wire conference circuits. A national line connects the Warning Centers to the Regional Warning Centers, eight regional loops connect the Regional Warning Centers to states and other federal warning points, and 48 state circuits connect warning points within each state. Each of these circuits or loops is essentially a party-line or conference line. If you pick up one of the phones, you'll hear if someone is talking on any of the others.

Nationwide messages generally originate in the National Warning Center or its alternate. Historically, the National Warning Center was at Cheyenne Mountain, Colorado (with NORAD where alerts would most likely originate), and the Alternate National Warning Center was at Denton, Texas. There was a second alternate facility in 1968 at Olney, Maryland. At some point, FEMA seems to have "promoted" the sites to remove Cheyenne Mountain, making Denton the primary and Olney the alternate.

Denton and Olney were both the locations of FEMA regional headquarters, which have gone by various names over time like Special Facilities or Federal Support Centers. They were originally built by Civil Defense or FEMA (depending on the year) to coordinate the recovery from nuclear attack, and as such they were hardened. Perhaps the most famous is the Region 8 "FEMA Bunker" at the Denver Federal Center. Other FEMA regional headquarters tended to be either in more remote areas or were built by repurposing existing military facilities (such as that at Olney, a reused Nike missile site), and as such kept a lower profile.

It's hard to know what exactly is going on today. In the 2016 NAWAS Operating Manual, the most recent version that seems to have been made public, the Alternate Warning Center is directly named as Olney, MD but the location of the primary is left unsaid. The Olney facility has been transferred to the Naval Surface Warfare Center and is no longer in use by FEMA. A document suggests that the Alternate Warning Center may have moved to Thomasville Federal Center, near Atlanta. Denton, TX remains a major FEMA site and may still be the primary.

The primary control circuit connected the (originally) three warning centers with the eight regional headquarters, but considering that some of the regional headquarters were themselves warning centers, it had fewer points that it might sound like. That circuit served as one of several ways (including the HF radio system FNARS) that these major FEMA sites could communicate with each other in a major disaster, but it was less important for alert dissemination.

The more important parts are the regional circuits, which are configured as loop to provide redundancy against a line break. The regional circuits connect a FEMA Regional Headquarters, at least one Warning Center for redundancy, federal alerting points (an AT&T document lists Coast Guard stations as an example), and a primary and alternate warning point within each state. FEMA would use these circuits as the main way to distribute a nationwide alert, reading it directly to the state Warning Points, which are generally the offices of the state's emergency management agency. For example, in New Mexico, the Primary Warning Point is the Department of Public Safety office in Santa Fe; the Alternate Warning Point is the state Emergency Operations Complex at the National Guard complex by Santa Fe.

There are likely some additional levels of redundancy in most cases. For example, state emergency planning documents imply that the Department of Energy has a NAWAS site at Kirtland Air Force Base that is connected to at least the regional and state networks, and possibly the national control circuit as well. It thus serves as an additional contingency for distribution of alerts on the state network, were the Primary and Alternate Warning Points to be lost.

Each state circuit is left in part to the discretion of the state, with criteria stated by FEMA. Generally, it includes the state warning points along with county emergency operations centers and major infrastructure facilities like hospitals and power plants.

Along with the specialized purpose of NAWAS come specialized equipment. NAWAS relies on a system that AT&T calls SS1, presumably Signaling System 1, but is more similar to a selective calling scheme than to a more general signaling system. SS1 appears fairly similar to the control pulses used by CDWS, but repurposed to ring phones on NAWAS to get a user's attention. On the national control circuit, for example, the National Warning Centers have SS1 transmitters that can be used to signal SS1 receivers at the Regional Headquarters to ring. Keep in mind that this is conference system, so there is no real sense of "placing a call" or "hanging up." The telephones are always connected. The provision of selective calling is just to get another user's attention so that they pick up the phone. In practice, they likely won't even pick up the phone, as most NAWAS sets are equipped with an always-on speaker to monitor activity on the circuit.

The SS1 selective ringing system also allows the National Warning Centers to selectively call state warning points on the regional circuits. State warning points, at least those operated by the state itself, are able to selectively call other sites on the state network. And that's actually nearly the limit of the SS1 capabilities. All other selective calling is done by "voice paging," basically yelling into the phone in the hopes that the party you want is listening to their speaker.

There is one other interesting capability of SS1, which requires understanding the structure of the network, beyond just the circuits. At each state primary warning point, the regional NAWAS circuit is actually bridged to the state warning circuit, so that any traffic on the regional circuit will also be heard on the state circuit. Essentially, in its "normal" state, the whole state circuit is just a leg of the regional circuit, and all sites on it hear regional traffic. This ensures that a warning read by one of the National Warning Points will be heard as quickly as possible through the over 2,200 phones in the total NAWAS system. This connection is only present on one pair, though, so it's one way: the state circuit hears traffic on the regional circuit, but the regional circuit does not normally hear traffic on the state circuit.

As originally designed, a foot switch in each state Primary Warning Point disconnects the two networks when depressed, allowing the Primary Warning Point to "speak" on the state circuit only. The foot switch basically selects which of the two networks the Primary Warning Point phone will transmit onto, but also disconnects the bridge so that the Primary Warning Point can speak on the state network even if the regional network is busy. You can imagine, though, that this would pose a problem when a critical alert needs to be distributed. To guard against a stuck footswitch or just a particularly chatty state emergency manager, the National Warning Centers can send an SS1 code that will throw a relay to bypass the footswitch and reconnect the regional and state networks. This code would be sent just before any critical nationwide warning.

The use of four-wire conference systems usually requires a slightly different type of telephone set, anyway. One of the goals of four-wire systems is often to function as a "squawk box" or "hotline" (one of the many definitions of the word), an always-on system that can be heard on a speaker at every connected location. As a result, every NAWAS warning point has a loudspeaker. There is also a telephone handset, used to speak into the system. Because there are many phones on each NAWAS circuit and you know how large Zoom calls tend to get, each NAWAS handset has a push-to-talk button. You don't see these very often today, but Western Electric offered telephone handsets with a PTT button mounted on the inside of the handle as a standard product. Apparently depending on the preference of the installation site, the loudspeaker is automatically disabled either when the handset is picked up or only when the PTT button is depressed. This prevents feedback or echo. It might seem a little odd to hear a conversation on the speaker but speak into a handset, but this was very common in mid-century telephony, as the echo and feedback problems of a true "speakerphone" proved hard to solve.

At State Warning Points, an indicator light attached to the telephone set helps the operator understand what circuit they are connected to. Green means that the footswitch has been depressed to disconnect the state network from the regional network. White means that another location, presumably the State Alternate Warning Point, has depressed their footswitch to disconnect the two systems (the bridge was duplicated at the two sites, but one site pressing their footswitch would disconnect the bridge at the other site as well). A red light indicated that the National Warning Center had bypassed the footswitches to distribute an alert.

BSP 310-530-901 LL Issue A (1968) provides a general description and operating information on NAWAS as originally built. Much of the BSP relates to the bureaucracy of operating a special service on a nationwide telephone system: it lists, for example, the specific plant control offices responsible for the testing and maintenance of each circuit. The Cheyenne Mountain Central Office must have been an interesting place, with primary responsibility to handle trouble reports and network management on the national and regional circuits. Each regional circuit was assigned to a CO within the region as well, typically one in a major city near the corresponding FEMA regional headquarters.

These plant control offices are informed that they must report all outages and work done on the circuits to OCD, and that they must take special precautions to ensure uptime on the NAWAS circuits. 8.05 states, in characteristically dry AT&T tone, that "the circuit may be required by the customer at a moment's notice due to the nature of the business for which it is used." I have heard anecdotes that switchmen working in central offices were sometimes used to hearing routine chatter on the NAWAS, and I see that the BSP suggests (but does not require) that control offices responsible for the system have a dedicated monitoring speaker. Proving that customer service is always the hard part, the BSP requires that these control offices make regular visits to customer sites and emphasize the importance of keeping the loudspeaker volume turned up, using the PTT button on the handset, and promptly reporting any trouble.

The BSP also lays out a process for routine maintenance on this critical, high-uptime system, which AT&T refers to as "circuit line-up." Such line-ups are performed only on Saturday, and must be authorized by the National Warning Center and announced to all Warning Points. Once started, the line-up time allows the test rooms at each control office to perform routine quality measurements on the circuits and detect any problems requiring maintenance. The test room must keep the circuit audible on a speaker for the duration, so that they can immediately stop their testing should the National Warning Center transmit that they need the circuit returned to carry emergency traffic. Following each line-up, the National Warning Center would place test calls to each State Warning Point to verify correct functioning, and the test room was expected to stay on the line to monitor this process. One can imagine that the whole thing felt like a hassle to the test room personnel who had to stay perpetually on edge.

The modern Operations Manual for NAWAS is available via FOIA, or at least a version from 2016. It mostly describes the details of operating the station equipment, which have not changed all that much since the 1968 BSP. The original Western Electric sets have been replaced by NAWAS terminals manufactured by Comlabs, a small company focused on emergency communications. They appear to be modified AT&T or Lucent phones with a new dialpad (including the elusive DMTF ABCD "digits"), an alert light, speaker volume knob, and presumably the electronics swapped out for four-wire operation. Phones used in local warning points have no keypad at all, since they are not intended to issue alerts. The pulse-based SS1 system has been replaced by DTMF; selective calling is done using four-digit numbers while commands like linking and unlinking regional and state networks use the DTMF "A" digit.

Besides some modernization of the terminals, the sites on the system have expanded along with its scope. NWS forecast offices, NOAA facilities like the Hurricane Center, and FEMA's "MERS" mobile response teams have been added to the regional circuits. The NWS runs its own round of tests to ensure it can distribute weather alerts to State Warning Points.

You might be somewhat familiar with NAWAS. I have mentioned it before, and in general, it's one of the best known of the government's various emergency communications system. No small part of this is due to a compelling bit of drama available in the documents. The manual has scripts.

This is the FEMA (Alternate) Operations Center. A nuclear weapon detonated in (city, county, state) at _____ Zulu. Radioactive fallout is possible! Persons in (city, county, state) should be advised to remain under cover and await further instructions from state or local authorities. Residents are advised to take protective actions in accordance with local community shelter plans and to be alert for further instructions from state or local authorities. Residents in all other areas are advised that protective action is not required at this time.

This is actually somewhat lengthy considering that the audience for the message is mostly state emergency management agencies. Alerts of this type on NAWAS would be proceeded by an audible alert tone to attract attention, and followed by a roll-call of all state warning points to ensure that the message was received. Local authorities would do... something. In the event of an attack, they will theoretically activate local sirens with a wavering tone for 3-5 minutes. FEMA procedures dictate that this signal exclusively indicates a confirmed impending or in-progress attack. In practice, it is either the same as or not readily distinguishable from the signal used for tornadoes in many tornado-prone areas of the country---generally the only areas with sirens at all. FEMA procedures have often coped poorly with the use of warning systems for purposes other than those that were apparent during the Cold War.

The manual provides scripts for other scenarios, ranging from a detection by NORAD (probably actually by intelligence community assets) of possible fires to to reentering space debris, errant weapon launches (nuclear or conventional), and various natural disasters. One of the newer additions to the procedure is use by the Pacific Tsunami Warning Center to disperse warnings of tsunamis affecting the west coast. This is one of the newest disaster scenarios to prompt serious investment in mass notification, and large parts of the west coast are now equipped with sirens in case of tsunami.

NAWAS has a close connection to IPAWS, the Integrated Public Alert and Warning System. Most NAWAS alerts would simultaneously be issued as IPAWS alerts, distributed across radio and television stations and via the newer Wireless Emergency Alert system. Much like IPAWS, only federal authorities can issue national NAWAS alerts, but regional authorities like state governors are permitted and even encouraged to use state NAWAS circuits to disseminate local alerts. Indeed, most state governors residences have NAWAS terminals installed for ready access.

NAWAS has provided reliable service for many decades, but now shows its age. Private-line telephone systems like NAWAS are fundamentally challenging to harden against attack and disaster due to their fixed routing. Besides, the statewide conference line capability of NAWAS now seems rather limited compared to the popularity of text messaging in emergency management. In 2022, FEMA awarded AT&T a $167 million contract to modernize multiple FEMA communications systems, including NAWAS. The plans for NAWAS are vague:

AT&T will transition the NAWAS legacy technologies to newer services available via EIS through a well-planned, phased, cost-effective, and non-disruptive approach to the new solution with government oversight.

Sounds great. Can't wait for the new solution.

So there we have it, a large, nationwide communications network made up of four-wire private lines. There have actually been a number of these used over time, including notably the Strategic Air Command's C2 network which involved both radio links and private lines terminating at telco-furnished turrets at SAC bases. Some smaller-scale four-wire systems remain in use today for local emergency management purposes.

Later, possibly as soon as next week, we will take a look at a much more complex type of private line service: Common Carrier Switching Arrangements. But I might get sidetracked on the road to CCSAs once again, and post first about the federal private line services that lead to the invention of the CCSA: AUTOVON and the Federal Telephone System (FTS).