[ Home | $$Order$$ | Email to us ]

Cellular Networking Perspectives

David Crowe’s Wireless Telecom Magazine Articles

Q3‘2006 Issue

Emergency Alerts

Floods; Landslides and Avalanches; Forest Fires; Earthquakes; Hurricanes; Tornadoes; Tsunamis; Child Abductions; Dangerous Criminals on the Loose; Terrorist Attacks. There is fear in the air and, with an increasingly mobile population, a fear that people will not be listening to the radio or watching the television when important emergency notification and response information is broadcast.

For decades, radio and television stations have broadcast emergency information in Canada, mostly in response to natural disasters. Today, in our mobile society, using cellphones to reach more people with emergency information seems like a necessity.

The CWTA formed a committee to study public alerting in September, 2005 and will be basing its work on earlier EAS (Emergency Alert Service) work in the United States as well as specific Canadian requirements, technologies and laws. Coordination with the United States and Europe, which use many of the same wireless technologies, will be particularly important if standards need to be modified.

Information provided to the public should include the type of event, the area it will affect, the estimated time (if it is a predicted future event) and recommended action to take. It can then be used by Canadians to take action to protect their family, neighbours and friends. Equipment and buildings can be prepared. Areas near the disaster can be avoided and people in those areas can leave, if necessary.

There are two media that can be used to provide this information to cellular phones: voice and text messaging. If text messaging is used, mobiles can be alerted individually (Point-to-Point SMS, or just ‘SMS’) or as a group (Cell Broadcast or CBS).

There are significant differences between types of events, and this affects the characteristics of the needed communications. Most emergency communications, unfortunately, are time critical, and any communications must occur within a few minutes. A few types, such as reports of abducted children or escaped criminals, can be spread out over a longer period of time.

Community Voice Alerts: Reverse 911

Many police and community organizations have the ability to do an automated call-out to a large number of phones. The phone rings and, after it is answered, a recorded message is played. In some systems, to avoid messages being left on voice mail systems, a key must be pressed on the phone to indicate that a real human being is listening. Without acknowledgement the call will be attempted again later.

Wired phones are grouped geographically, meaning that, with the cooperation of the phone company, reaching everyone with such a phone in a specific area is quite easy. With cellular phones it is more difficult. However, given that every MSC (Mobile Switching Center) must maintain a list of phones that are in its coverage area, it would be possible to design a system that would also ring every cellular phone that is currently in the coverage area of a specific set of base stations.

The biggest problem with this technique is that it requires a lot of communications resources, and consequently it is not suitable for true emergencies. In an emergency situation, the communications system would already be overloaded, and a flood of calls telling people about the emergency could easily overwhelm the system, making it difficult for people to dial out, and possible crashing the entire system. Many people will have already learned about the emergency in other ways, and preventing them from making important calls is not a good tradeoff.

Telus is conducting field trials of a system in Ottawa, Vancouver and Coquitlam that is more precise, reducing the use of system resources as well as reducing notifications to unaffected people. This is based on a GIS system that uses the 911 address database to determine whether each phone is within the emergency zone. New Brunswick tested a similar system in 2004 that was designed to make 5,000 calls per hour. Others are examining whether the internet could be used to distribute alerts instead of the phone system.

In May, 2006 Bell Canada announced the availability of wireless Amber Alerts (notifying the public when a child has been abducted) by subscription, one of the first wireless companies to provide this service.

Community voice and internet alerts will be increasingly important for this public service and for use by the police to warn people about an increase in criminal activity, or the suspected presence of a dangerous criminal. However, they are not a satisfactory solution for natural disasters or terrorist attacks.

Getting the Point Across with Point-to-Point Texting

Almost all cellular phones have basic text messaging (SMS) capabilities, and can receive several messages limited to 140-160 characters in length. To deliver the same number of notifications, text messaging takes fewer system resources than voice, and the messages can be successfully delivered even when someone is in a call or otherwise unable to answer. The only phones that will not support SMS are analog phones, phones being used in an analog system and extremely early models of digital phones.

Normal text messaging is described as “Point to Point” because each message is destined to a single mobile phone. Often referred to as texting, this accounts for the vast majority of text messaging traffic today. Because it requires one message for each mobile, it is not a desirable solution for the worst emergencies as it can overwhelm an already overloaded system.

System loading is not as bad as for Reverse 911 because trunks between systems do not need to be reserved for these calls and a message can be delivered in a fraction of a second, rather than the many seconds needed to speak a recorded message. Future optimizations might mean that only one message is sent to each MSC from the emergency message originator, reducing backbone network traffic dramatically, leaving the MSC to duplicate the message for each mobile in its coverage area. The MSC and associated base stations could also manage the loading intelligently, sending the text messages only when capacity exists on the system. However, there is a limit to the number of short messages that can be delivered over the radio interface, even if network traffic was reduced to an insignificant level. And, none of these optimizations exist today.

Current systems cannot guarantee delivery in a reasonable time. Cingular Wireless, in a briefing to the US FCC, claimed that it could take about 2 hours to notify everyone in an average-sized city with a small text message.

The biggest problem for SMS emergency alerts is the littleness of the messages. Text messages for both GSM and CDMA (as defined by 3GPP TS 23.040) are generally restricted to 140 characters (fewer than there are before “140” in this paragraph). It is true that long messages can be broken into several segments, however this not only significantly increases the delivery time, but is unreliable because “the concatenation method [of all the segments belong to one message] is unsophisticated and provides no means of requesting missing message segment units” (ETSI TR 102.444). For this reason it is not recommended to send a message that requires more than 5 segments, a total of about 700 characters.

In the United States it is recommended that spoken emergency warnings be no more than 2 minutes in length. Assuming a relatively slow rate of speaking of 150 words per minute, and an average of 6 letters per word (including a space between words) 1800 characters would be needed.

Figure 1 shows a recent Canadian severe weather warning. With only limited information about precautions that affected people should take, it includes 957 characters, significantly over the recommended limit of 700 characters in a text message segmented into 5 pieces.

Figure 1: Sample Canadian Weather Warning

 

1:22 PM EDT Monday 19 June 2006

Severe thunderstorm warning for Apica Mountain area continued

A line of strong thunderstorms stretching between Bagotville and the Parc de la Mauricie is moving slowly eastward while intensifying.

Another one at the western Abitibi limit is moving towards Rouyn and Ville-Marie at 50 km/h. These thunderstorms will produce gusts of 90 km/h or more – hail of 2 cm or more – heavy rain – and frequent lightning.

Public in these regions should take the necessary precautions and listen for subsequent severe weather warnings.

This warning is in effect from 01:00 PM to 03:01 PM EDT. A line of strong thunderstorms stretching between Bagotville and the Parc de la Mauricie is moving slowly eastward while intensifying. Another one at the western Abitibi limit is moving towards Rouyn and Ville-Marie at 50 km/h. These thunderstorms will produce gusts of 90 km/h or more – hail of 2 cm or more – heavy rain – and frequent lightning.

 

This, by SMS standards, is not a small message. Cingular’s analysis estimates that even if a message like this was trimmed to fit in 5 segments, it would still take about 10 hours to deliver to everyone with a cellphone in an average city!

Another important caveat is that correct handling of segmented messages has not been validated for all phone models. Since commercial text messaging services generally do not allow messages large enough to require segmentation, there is little practical experience with this capability.

Security is also a concern for SMS. There is no indication that a message is generated by a legitimate authority that cannot be emulated by typing in a text message from another phone. It is true that the phone number that the message appears to be from cannot easily be spoofed, but since this may have to be different for each emergency information provider, it is unlikely that most phone users will be able to tell whether the number is valid or not.

Point to Point text messaging is suitable for community alerts, such as abducted children and escaped felons. These messages can be kept under 160 characters so they fit in a single message segment. The good news is that, if consumers can opt in or out, the number of messages may be significantly less than the number of phones. This method is not a good solution for life-threatening emergencies where everyone should be notified within a few minutes.

Cell Broadcast: Text Messaging a Group at a Time

The system loading created by attempting to SMS every mobile has led to increasing interest in Cell Broadcast, the ability to send a single text message to every idle mobile in the coverage area of a cellsite. This significantly reduces the capacity problem, as the number of messages is divided by the average number of mobiles registered in a cellsite. Although it is not specified by today’s standards, mobiles engaged in a call could conceivably be messaged by point-to-point SMS, extending the reach to all mobiles in good coverage areas, except those in the process of setting up or tearing down a call, or in other intermediate states where message reception is not possible.

Some people have leapt to the conclusion that these characteristics make Cell Broadcast the Holy Grail of emergency alerting. Sadly, however, this is far from the truth.

Cell broadcast shares many of the limitations of point-to-point SMS, including, with some technologies, even more stringent limitations on the length of messages (e.g. 88 byte/93 character segments). This makes it even more difficult to transmit detailed messages.

The loss of message segments is a bigger problem because there is no way to verify that the message made it across the radio interface (the most likely place for a message to be lost) and reached all mobiles. Compensation for this problem is usually through the retransmission of the message at regular intervals throughout the emergency. This, however, requires that mobiles correctly filter out the duplicates to avoid displaying the same message several times.

A cascade of problems are caused by the detrimental impact of cell broadcast on battery life. GSM experts have reported that standby time can be reduced up to 50% because of the need to monitor an additional control channel. Not only will this mean that some phones are unavailable at the time of an emergency due to a dead battery, but it will lead many people to turn off cell broadcast. In fact, many phone manuals recommend this, and many phones come with cell broadcast turned off.

The cell broadcast channel allows messages to be categorized, and only a few of those categories will be for emergencies. The numbers of these categories have not been standardized, so nobody knows which cell broadcast sub-channels should be permanently turned on in phones and the phone cannot determine which sub-channels require special handling.

There are no requirements for the cell broadcast user interface, so one implementation may be to merely add the cell broadcast to the end of the list of incoming, unread text messages, putting “Tornado heading for downtown” just behind “cu@8” and “pls get bread on the way home”.

One good thing about cell broadcast is security. There is no way for an outsider to generate a cell broadcast message, so false emergency alerts are unlikely using this method. However, unless phones visually discriminate between cell broadcast and point to point SMS, the average user would be unable to tell the difference and a point to point SMS might be able to appear like a cell broadcast on some phone models.

There has been little commercial interest in cell broadcast, the application that the technology was developed for. This is at least partly because there is no way to bill for the service. Without acknowledgments of the messages (which is difficult in principle for a broadcast service, and not implemented at all for this particular service), there is no way to know whether a mobile received a particular message. Consequently there is very little equipment available to support cell broadcast in the network, including routing, store-and-forward and testing.

It is also not known how many cell phone models support cell broadcast correctly, e.g. whether they suppress duplicate messages and correctly splice segmented messages together.

The safest practice in the short term would be to limit each broadcast alert to a single segment (e.g. 88 characters), limiting the service to alerting people of the type of emergency and encouraging them to tune to a radio or television station or visit a website (with a compact URL) to get updates.

The Role of Location Technology

Location technology might not appear to have a role in emergency alert systems, as messages will be delivered only to base stations with a coverage area overlapping the area of the emergency. The key word is ‘overlapping’. Alerts may be delivered to a large cellsite that has only a small overlap with the area of the emergency. This occurs because the zone in which each alert should be delivered is defined using administrative areas, whereas base stations are defined by a combination of the laws of physics, the efforts of RF engineers and a variety of other factors, some of which vary with the seasons and the time of day.

It is important to minimize the number of alerts people receive, or “Boy Who Cried Wolf” fatigue sets in, with people not taking life threatening situations seriously as they have been interrupted by many irrelevant alerts in the past.

GPS and other location technologies must be used carefully, however. If phones have to access the network to get an update on their position, this will just increase system loading at the worst possible time. However, as GPS technology becomes more sophisticated, there will be more phones that are always aware of their location. These phones can safely ignore alerts for many emergencies that affect only a small area.

Another way to reduce message fatigue is to reserve widespread alerting for truly life-threatening situations. Other alerts, such as for missing children or non-life threatening weather conditions, should be limited to people who opt-in to specific services.

Other Creative Ideas

The goal of getting emergency information to the people has resulted in some creative ideas. One of them is to merge a weather radio, where many emergency alerts are already broadcast, with a cellular phone. The weather radio would share the battery, display, keypad and computer/memory with the cellular portion of the phone. The battery drain is minimal, but the radio module and large antenna would make the phone too unwieldy for some.

A similar idea is to provide an FM radio in the handset which could be tuned to an emergency frequency when not in use. If the FM radio was being listened to, it can be assumed that the radio station to which it is tuned will provide the emergency information. The antenna can be built into the headset, but this would require a special headset, and the phone could not receive emergency alerts without it.

Broadcast video is also coming to some cellular phones and similar wireless devices. When this achieves significant market penetration, it may become a useful adjunct service, providing maps and photographs, as well as text.

A New Design

I have thought for quite some time that a new design is required to really make emergency alerting work on cellular phones. Many of the pieces are in place, but they need to be put together in creative ways. Modifications will be required to phones and to cellular network equipment. A new network will be required to link emergency information providers to cellular networks.

Phones will have to recognize life-threatening emergency alerts and alert the user even if all sounds on the phone have been turned on. The message should be displayed on the screen in the most visible way possible, and never queued behind other messages of lower priority. Standardized categories for messages must be defined, and these should be the same as those used in the United States, Europe and the rest of the world (although some reservations for national use only can be made).

Network standards must be designed so that equipment, such as MSCs and Base Stations, can deliver alerts efficiently and with priority, based on the area that the notification is supposed to cover. Transmission should be by a combination of cell broadcast and SMS. Changes may not be required at the lower protocol levels, but a new application level protocol, at the very least, will be required.

The alerts will be injected by a secure network (yet to be developed) that will give access to the many local, provincial and federal agencies that may have the need to generate alerts, while preventing access by hackers and troublemakers. This network will have to include the logic to allow the efficient distribution of messages to cellular voice, text and packet-data entities in the network.

My concept of a future network is illustrated in Figure 2.

Figure 2: A Future Public Alert Network

Thoughts for the Future

As Cingular, the largest US cellular carrier, recently wrote to the Federal Communications Commission, “The ‘silver bullet’ for supporting a comprehensive Wireless Emergency Alert Service is currently undefined.”

Emergency alerting via cellphones will become an important supplement to existing emergency response communications technologies. Existing technology has critical deficiencies for this purpose, but if used carefully, it could still inform thousands of people who would otherwise be unaware of an emergency situation. Careful design of future systems could extend the capabilities greatly without adding much complexity or cost to phones or cellular systems. The silver bullet may be within our grasp if all groups get together and act cooperatively, thoughtfully and creatively.

Table 1: Emergency Alerting Resources

ID

Name

BCE AA

Amber Alerts. Bell Canada Enterprises. www.bce.ca/en/amberalert/overview/index.php

CANALERT

strategis.ic.gc.ca/epic/internet/inet-tdu.nsf/en/h_wj00019e.html

Cingular

Briefing on Wireless Emergency Alert Service. June, 2006.

CWTA

White Paper on Wireless Public Alerting in Canada. CWTA. February, 2006.

Maps

Atlas of Natural Hazards in Canada

atlas.nrcan.gc.ca/site/english/maps/environment/naturalhazards

NB GPAS

New Brunswick Geographical Public Alerting System. March, 2004. strategis.ic.gc.ca/epic/internet/inet-tdu.nsf/vwapj/SummaryReportMay14.pdf/$FILE/SummaryReportMay14.pdf

TR 102.182

ETSI TR 102.182. Emergency Communications (EMTEL); Requirements for communications from authorities/organizations to the citizens during emergencies. webapp.etsi.org/action/PU/20060321/tr_102182v010101p.pdf

TR 102.444

ETSI TR 102.444. Analysis of the Short Message Service (SMS) and Cell Broadcast Service (CBS) for Emergency Messaging applications. February, 2006.

TS 23.040

3GPP TS 23.040. Technical realization of the Short Message Service (SMS). March, 2006. www.3gpp.org/ftp/Specs/html-info/23040.htm

TS 23.041

3GPP TS 23.041. Technical realization of the Cell Broadcast Service (CBS). March, 2006. www.3gpp.org/ftp/Specs/html-info/23041.htm

TS 44.012

3GPP TS 44.012. Short Message Service Cell Broadcast (SMSCB) support on the mobile radio interface. February, 2005. www.3gpp.org/ftp/Specs/html-info/44012.htm

US FCC EAS

The Emergency Alert System (EAS). www.fcc.gov/cgb/consumerfacts/eas.html

Weather

Public Weather Warnings for Canada. Environment Canada. www.weatheroffice.ec.gc.ca/warnings/warnings_e.html

About the Author

David Crowe is a wireless standards, technology and numbering resource consultant based in Calgary. He can be reached at David.Crowe@cnp-wireless.com.

  Comments

Your name:
Your email address:
   

© – Copyright Mon, May 14, 2007: Cellular Networking Perspectives Ltd.