Several recent innovations or capabilities under development provide some of the necessary ingredients for an end-to-end, all-hazards warning system that fully exploits geographical information:
• The Common Alerting Protocol (CAP) standard for formatting alerts includes geographical locations by Federal Information Processing System (FIPS) code or vertices of polygons to define affected regions along with information about the source and nature of the alert and the action to be taken.
• Cellular phones and other mobile devices “know” where they are located (at a minimum using mandated E911 location capabilities and, increasingly, using embedded Global Positioning System (GPS) receivers and other location information such as nearby wireless access [Wi-Fi] sites) and increasingly possess considerable processing power, high-resolution displays, and the like. More generally, computing devices, such as laptops, desktops, and cable set top boxes, either can establish their location or can easily be outfitted to determine such information using one or more of the approaches listed above. Wired devices can also use knowledge about the physical location of the networks to which they are attached to establish their location. Applications either built in or installed on these devices can be used to receive and present targeted alerts and warnings. Importantly, even if the systems designed to transmit alerts/ warnings cannot precisely send messages only to the desired set of recipients, the receiving device can use knowledge of its location, together with
geographical information coded in the message, to deliver messages only to someone at the specified location.
• Tools for geotargeting at various resolutions are becoming increasingly available, and these have been adopted by the advertising industry. The alerting community may be able to adopt or adapt capabilities being developed and used for advertising.
Much of the second day of the workshop focused on new and emerging technologies and tools for determining location and disseminating geotargeted alerts and warnings.
Although discussion at the workshop tended to focus on new technologies, particularly mobile devices, several presentations examined how older technologies can be used in ways that provide enhanced geotargeting capabilities. Rick Wimberly, Galain Solutions, examined reverse-dialing alerts; John Kean, NPR Labs, discussed innovations in radio broadcast; Bruce Thomas, Midland Radio Corporation, examined weather radios; and Ron Boyer, Boyer Broadband, discussed alerting over cable television systems.
Reverse-dialing alerts allow for officials to auto-dial landlines, or mobile numbers of registered users, within a certain area and play a prerecorded alert. As noted by workshop participant Ken Rudnicki, City of Fairfax, Virginia, reverse-dialing systems are currently one of the better tools emergency managers have to provide geotargeted alerting. However, the system still has several challenges.1 The challenges, as discussed by Rick Wimberly, include the following:
• When large sets of numbers are dialed, this can overwhelm local phone switches and cause calls to be dropped.
• The significant decrease in the number of households with landlines reduces the reach of these systems. Reverse-dialing systems can
1 In a 2012 article, Rick Wimberly examined the shortcoming of these systems during the 2012 Colorado wildfire where approximately 25,000 of the 118,000 reverse-dialing alerts were not delivered. See R. Wimberly, Flawed delivery: Do alert notifications fail to live up to expectations, Emergency Management Magazine 5(7):22-35, 2012.
reach mobile devices, but this requires subscribers to register their phone numbers. Despite communities aggressively encouraging people to register, registration rates for mobile subscribers across the country are still well below 10 percent.
• Reverse-dialing systems are not particularly effective at delivering messages to those with disabilities.
• Reverse-dialing systems are expensive, and local jurisdictions may not be in a position to purchase or modernize a system.
Radio Broadcast Technologies
NPR Labs, a small, self-supported broadcast technology research and development outfit operated by National Public Radio, is currently examining the use of two new technologies that may benefit alerting: broadcast repeaters and the use of the radio broadcast system (RBDS).
NPR Labs partnered with Geo-Broadcast Solutions (GBS) to examine the performance and use of GBS technologies known as ZoneCasting and MaxCasting. In both technologies, a group of synchronous repeaters repeats the signal of the primary station using lower power and transmitter heights. In MaxCasting, the nodes are time-aligned to the primary transmitter to reinforce or extend coverage. In ZoneCasting, the individual nodes can be used to send distinct programming to different locations. Figure 2.1 demonstrates how these tools can expand coverage of a radio station and also provide separate coverage by zone.
John Kean discussed how both tools support alerting: first, they extend the reach of radio alerts to communities currently poorly served by single radio transmitters; and, second, by supporting distinct programming by different nodes, they enable geotargeting of alert and warning messages. Although they require new equipment on the part of the broadcaster, they have the advantage of requiring no new equipment for the public.
NPR Labs is also working to demonstrate the use of RBDS to reach at-risk populations, including those with hearing impairments. RBDS is a standard to embed small amounts of digital information in conventional radio broadcasts that almost all FM stations are capable of supporting. It is currently used most often to transmit and display song or other program information and is commonly found in automobile radios. One of the objectives of the NPR Lab project is to experiment with using RBDS to send text information using household receivers to people with hearing impairments to explore how effectively this technology would reach this large segment of the public.
NOAA Weather Radio
The National Oceanic and Atmospheric Administration (NOAA) Weather Radio (NWR) was originally developed in the 1950s and 1960s to provide weather observations and forecasts to those in flight or at sea. In 1974, an outbreak of 146 tornadoes within a single day spurred the expansion of the service. Bruce Thomas noted that today, more than 1,000 broadcast transmitters provide coverage across the United States and its territories.
NWR uses the Specific Area Message Encoding (SAME) standard to geotarget its alerts. Adopted in 1988, SAME was the first geotargeted alerting standard and remains an important foundation for geotargeted alerts to this day. SAME allows NWR to target at the FIPS code level. This generally means at the county level (or equivalent geographical area); however, large cities located within counties may have their own unique FIPS and SAME codes. Additionally, high-risk areas may have a unique SAME, such as areas around a nuclear power plant. More precise geotargeting may be possible by adjusting the first digit of the SAME code.
Boyer discussed current capabilities and opportunities for enhanced alerting over cable networks (more formally known as multichannel video programming distributors or MVPDs). Currently, cable providers are required to distribute presidential alerts, test their alert system weekly, and monitor two Emergency Alert System (EAS) origination sources. Many operators support additional alerting capabilities on a voluntary basis.
Today, an alert received from EAS is distributed to all the subscribers to a cable system, even though they may live outside the specific region that is the subject of the alert. In principle, modifying cable boxes to know their location and filter messages accordingly could provide better geotargeting. Boyer explained that this is not entirely straightforward since cable boxes have been designed chiefly to decode video content, were not designed to be location-aware, and the systems that would be needed to link a subscriber’s cable box to the subscriber’s address in the MVPD’s billing or operational systems do not exist today.
A second option would be to determine location via service nodes within the cable system. However, these do not necessary follow the geopolitical boundaries commonly used to geotarget alerts. Boyer noted that, as a result, adding enhanced geotargeting to MVPDs will most likely require enhancements to the entire networks, not just a single network element.
Although television is still the primary source by which the public receives information about disasters, this is rapidly changing as individuals spend more time using the Internet for infotainment. Richard Barnes, BBN Technologies, discussed the challenges of alerting over the Internet, and Hisham Kassab, MobiLaps, discussed how alerts and warnings can be introduced into streaming video content.
In principle, alerting over the Internet appears to be a straightforward task requiring, essentially, the delivery of a suitably formatted document that the Internet-connected device can render, something that is done trillions of times a day. However, this would require that the location of the Internet-connected device can be established by some combination of the device itself and the network and that a device either monitors for alert information or the alert can be placed in an information source that is already being monitored by the device.
Geotargeting Using Internet Protocol
Richard Barnes explained that geolocating based on Internet protocol (IP) address is very limited and generally relies on privately managed databases that match IP addresses to physical addresses. IPV6 has extra space, a 128-bit address, and many hoped that this extra space could be used to insert geolocation information and allow for better IP topology-based geolocation. However, IP addresses are assigned based on the network topology, not physical geography. In addition, IPV6 might make IP-based geolocation more difficult because the larger address space may make it more difficult to complete network traces used to determine location. Essentially, IPV6 has the same geotargeting capabilities as IPV4, where geotargeting using network tracing can be done in a metro area. (Law enforcement can obtain additional information on a geolocation of an IP address from a service provider with a subpoena.)
Another method for geotargeting would be to incorporate alert and geographical information in an information source that a user already frequently monitors. For example, an alert could be sent to Facebook, which would then send the alert to all of its subscribers in Nebraska. This bounds delivery of alerts to specific channels, and these platforms
have their own challenges in accurately geolocating their subscribers and geotargeting messages.2
Alerting Over Streaming Video
Increasingly, Hisham Kassab noted, Internet-delivered services, such as Hulu Plus, Netflix, and YouTube, that stream video over the Internet are starting to be used in place of traditional broadcast and cable television for which alerting systems already exist. There are a variety of ways to deliver and display alerts on Internet-connected devices like computers, tablets, or game consoles, including streaming video services and the applications that display them. With the right modifications, these services can be used to receive and display geotargeted alerts delivered or triggered by the video stream and displayed by the application used to view the content.
There are four steps to streaming video, content creation (e.g., Warner Brothers films a television episode), content provision (e.g., Hulu licenses content and makes it available to its subscribers), content transmission (e.g., the viewer streams content over a Comcast broadband connection), and content presentation (an application running on a device displays the content). It obviously is not practical to insert alert information at the content creation step, but alerts could potentially be inserted at any of the other three steps, each with a differing capability to geotarget alerts:
• Content provider. Two possible mechanisms could be used to geotarget at the content provider level: the billing address, as a proxy for location, or an IP address. Both of these have limited accuracy because a person may be accessing content away from their home, and IP address databases are not always accurate, as discussed in the previous section.
• Content transmission. ISPs may be able to use their knowledge about network topography to determine the physical location of their users.
• Content presentation. End-user devices, most notably tablets and smart phones, often have some information on their location. An alert could be inserted into the video stream, and once the alert reaches the device, an application could use its location to determine if the information is relevant.
2 For discussion of the challenges of alerting over Facebook and other social media platforms, see National Research Council, Public Response to Alerts and Warnings Using Social Media: Report of a Workshop on Current Knowledge and Research Gaps, The National Academies Press, Washington, D.C., 2013.
There are several ways in which a mobile user’s location might be determined. Farshid Alizadeh, Skyhook Wireless, Larry Dodds, TruePosition, and Ayman Naguib, Qualcomm, presented potential technologies for determining mobile device location.
Mobile Location Determination Using Wi-Fi Access Points
Traditionally, mobile device location has been established using two methods: GPS, which is fairly accurate but only works outside and takes significant time to obtain a location fix and cellular tower triangulation, which has comparatively poor accuracy but provides a faster location fix. Skyhook incorporates a third source, Wi-Fi access point signals. The technology works by matching access point and cell tower signals to a proprietary location database. The technology has a median accuracy of 20 to 40 meters and can return a location fix within a few seconds, and currently, Skyhook has almost 400 million access points in its database.
A unique challenge to using Wi-Fi access points is their dynamism. Wi-Fi access points are under varying people’s control and are dynamic—being moved, being removed, and being added. To compensate for this, Skyhook relies on a huge redundancy in these access points. Additionally, the location database is updated frequently. For example, when a user sends a snapshot of surrounding access points to Skyhook servers, not only does the system return a location, but it also incorporates those data into the database to calibrate the location of the access points.
Mobile Location Determination Using Television Broadcast Signals
TrueFix TV Positioning, developed by TruePosition, uses over-the-air (OTA) television broadcast signals to determine mobile device location. Similar to using Wi-Fi signal strength to determine location, these devices search for local broadcast signals and use the signal arrival times, which are directly proportional to the distance to the transmitter, to determine the location. A key benefit of this technology is that it can be used in indoor and urban environments where GPS signals are not able to penetrate building structures. According to Dodds, about 95 percent of the U.S. population lives in areas where the OTA coverage is sufficient to determine position. However, to fully use OTA for mobile device locating, a television-band receiver would need to be added to handsets.
Mobile Location Determination Using Uplink Time Difference
UpLink Time Difference of Arrival (U-TDOA), also developed by TruePosition, uses the time of arrival of signals at multiple cellular towers. Measurements are made using devices that are 1,000 times more sensitive than traditional base stations and are located at or near cellular towers. Several measurements are sent to a central node that calculates the location with fairly high accuracy, typically within a 50-m radius of the correct position. The technology is widely used today by cellular companies to provide E-911 services.
Indoor Geolocation of Mobile Devices
Indoor geolocation is more difficult because GPS and other signals used to determine location do not readily penetrate building structures, and because a significantly higher accuracy is needed indoors. Naguib noted that a 10-meter error while driving is barely noticeable, but it would be problematic for someone navigating through a building.
Qualcomm is developing an indoor positioning technique that uses three additional data sources to determine indoor positioning:
• Wi-Fi measurements;
• Building maps, which provide additional information on what locations are viable (places the receiver could not be located, such as within a wall) and routing information (transitions that are impossible, such as crossing through walls); and
• Sensors on the phone such as accelerometers, gyroscopes, and compasses that can be used as an inertial navigation system by which a prediction is made based on relative motion of the device from its previous position.
As described in Chapter 1, WEA provides limited capabilities for geotargeting alerts to mobile users. New technologies and innovations may provide additional capabilities for alerting and more narrowly defined geotargeting. George Percivall, Open Geospatial Consortium, discussed the use of Short Message Service (SMS) to report geotargeted information; J.T. Johnson, Weather Decision Technologies, described a third-party application for geotargeted weather alerts; and John Davis, Sprint, discussed possible approaches to enhancing the geotargeting capabilities of WEA.
Geotargeting of SMS
SMS is almost universally supported on mobile phones and widely used to send and receive text messages. George Percivall discussed the Open Geospatial Consortium’s (OGC’s) work to develop a standard, Open GeoSMS, for representing location information in SMS messages. The location can be displayed in a mapping tool or used to retrieve satellite images or other information about the location. Open GeoSMS was used in the mobile phone app Find Me Maybe, which was developed and deployed for limited use during Hurricane Sandy in 2012. The tool also subscribed to the FEMA SMS alert service.
Third-Party Application Capabilities
Using an application on a mobile device and location information from that device is another method for geotargeting alerts. One such application is iMap Weather Radio, which was developed by Weather Decision Technologies (WDT). iMap Weather Radio communicates NWS alerts to the public with the goal of providing some of the key features of NWR (e.g., always on and provides an alert tone that awakens users) on smartphones, which are much more widely deployed. J.T. Johnson noted that iMap Weather Radio offers the following features:
• Phone wakes up automatically with alerts;
• Alerts are for current location and for additional saved locations;
• Interactive maps provide radar view, alert polygons, and phone location; and
• Text and text-to-speech of alerts reaches drivers or those with visual impairments.
To improve accuracy and continuity, WDT uses triple redundant feeds from the NWS and clusters of redundant computers; the data center has an uptime of 99 percent. WDT has also begun working to incorporate CAP EAS into iMap Weather Radio by using the FM channel. WDT also works with local media so that an individual receiving an alert can then watch the local television news directly within the application, either as real-time content or prerecorded material.
Carrier Geotargeting of WEA
As discussed in Chapter 1, targeting methods used by carriers to deliver WEA vary. Geotargeting an alert to a small, defined area is the ultimate goal in the alerting community. John Davis highlighted chal-
lenges and opportunities for better geotargeting of mobile devices. These include the following:
• The current deployment of the long-term evolution (LTE)3 standard may provide a partial solution. With LTE, cellular IDs, which are used to determine if a tower is within the alerted area, are assigned to individual antennas on each tower, rather than the tower as a whole. This may allow for tighter sectioning of geographical regions within the tower’s signal.
• The use of GPS is a possibility for determining a mobile device’s location, but may pose a challenge if the GPS initiates a request to carrier networks to request location information. Davis noted that this may create congestion in the network and cause its failure. Furthermore, Davis reiterated that GPS inside buildings or on subways is a challenge, and commercial needs will drive the development of tools to allow this.
• Alerting systems may not need to do the geotargeting. A phone’s position may be determined by a combination of technologies, as described in previous sections, and then the phone can determine if an alert applies to its location.
• Another challenge is that geotargeting capabilities vary across carriers and devices and are an area of extreme competition between carriers. This creates barriers to discussions across organizations.
• The size of the message is probably the greatest hindrance. Davis explained that the biggest gain in encouraging appropriate public response, with the least impact on networks and devices, is the modest expansion of message information. An option to do this is to allow pagination of a message—that is, a series of messages that together provide the full alert text.
3 Long-term evolution is often marketed as 4G LTE.