Myanmar’s disaster: SMS or cell broadcasting for early warning?

May 19, 2008 (LBO) – The loss of tens of thousands of lives in Burma (Myanmar) in our region’s largest human disaster since the 2004 tsunami reminds us once again of the urgency of solving the problem of the last mile of the early-warning chain.

I no longer call these events “natural” because, for the most part, the deaths are caused by human action/inaction, not by the natural hazard per se.

This column is dedicated to the needlessly dead in Burma. I normally write about hard choices that do not allow one to keep the cake and eat it. But this column is an exception.

The answer to effective early warning is short message service (SMS) also known as texting AND cell broadcasting, deployed in parallel for different purposes.

The information of a powerful cyclone heading for the Irrawaddy delta was communicated well in time to the Myanmar authorities; the problem was that the warning did not complete the last mile to the citizens; the problem was that evacuation did not occur. The final few links of the chain were weak to non-existent. The chain broke at the weakest link, as chains do. People died, needlessly.

Citizens must be made aware through public warning. But more important is that first responders (persons in authority at the ground level, such as the Police or village leaders) must be communicated with BEFORE the public warning, so that they can organize orderly evacuations and conduct localized warning. Of necessity, media must be communicated with BEFORE the issuance of the public warning because they are the primary medium of public warning.

There are several technical solutions for communicating to first responders and the media: addressable satellite radio, dedicated radio systems separate from public networks, and SMS. Because redundancy is highly valued in the field of disaster risk reduction, there is no reason to place all eggs in one basket.

In the aftermath of the 2004 tsunami, LIRNEasia tested several of these technologies with a closed-user group of 32 Sarvodaya tsunami-affected villages .

In this column, I will discuss only one of these options, SMS to a closed-user group of media/first responders, and a very good public-warning technology that has not yet been fully tested in Sri Lanka, cell broadcasting.

SMS to first responders and the media

Analysis of the false warning and evacuation of the coastal areas of Sri Lanka on 12th September 2007 showed that the evacuation order was issued by political authorities without the full involvement of scientific personnel. One reason the scientific personnel could not play their rightful role in assessing the evidence was the lack of proper protocols and procedures within the various organizations.

The Department of Meteorology is designated the authority for receiving tsunami hazard information from international and regional sources, in addition to their role as the nodal point for cyclone hazard information. The Disaster Management Center (DMC) is designated as the authority for issuing public warning.

Therefore, the media should have been talking to the DMC and the Met Department should have been focused on interpreting the incoming hazard information and advising the DMC and the political authorities to the best of their scientific knowledge.

What happened was that the phones at the Met Department were fully occupied by journalists calling to get reliable information, preferably from the mouth of the head of the department himself. There was no time for scientific assessment of the incoming facts, including data from models showing landfall after three-four hours at least . The evacuation order was issued from Temple Trees.

One lesson to be learned from this mistake is that we must develop procedures to insulate the key scientific decision-makers from ad hoc news gathering at such critical times. They must be allowed to do their primary tasks in peace. The media must be provided with fast, reliable and trustworthy information that is not subject to misinterpretation (as would hurried phone calls).

The solution is a reliable, one-touch method of quickly alerting the news media, complemented by an updated website that will provide greater detail. The one-touch method has been developed and is now available for any government agency in Sri Lanka or elsewhere to use, subject to a few tweaks and tests that will be completed in a couple of weeks.

The SMS module that has been designed is part of the Sahana disaster management software suite , developed by Sri Lankan engineers and deployed throughout the world, from New York City to the Philippines. It allows government authorities to send a text message based on a pre-planned template to a group ranging from a handful to several hundred persons with one command. The numbers to which the message is sent are located in the Sahana database and can be used in the disaster relief and recovery phases as well.

One group of recipients can be pre-registered journalists; another can be the government first responders at the district level; another can be key officials and political authorities. They can be sent the same message or tailored messages.

As long as the button is pressed before word of the oncoming natural hazard reaches the public, it will be effective in quickly reaching the recipients. The system is not without weaknesses, but it is extremely low-cost and it is ready for deployment in weeks.

The principal weakness is inherent in SMS, which is a point-to-point communication technology (requiring switching to send messages to specific addresses). If too many SMS messages are sent that the same time (as happens after a bomb, for example), the network gets congested and the messages go through, but slowly. It took four hours for my SMS messages to get through on 26th December 2004.

It is like what would happen to the roads if all the motor vehicles in an area converge on them at the same time. The traffic would slow, if not stop altogether.

Theoretically, it is possible to build networks capable of carrying this kind of unusual-peak traffic, as it is possible to build roads that could carry all the vehicles in a region at the same time. Both would be horrendously wasteful, because the excess capacity would lie unused most of the time and every user will have to bear the extremely high costs. Of necessity, networks are designed to carry normal traffic, not extremely unusual peak loads.

Therefore, we have to take as given that all point-to-point networks (as opposed to broadcast or one-to-many networks) are prone to congestion. They are, therefore, unsuitable for public warning and coordination in the immediate aftermath of a disaster.

SMS is a point-to-point technology to which this conclusion applies. It is based on packet transmission and is therefore a little more robust that normal circuit-switched voice telephony, but it is nevertheless unsuited for public warning and immediate post-disaster coordination.

SMS is also limited in the number of characters it can carry in one message: 140 characters. This problem can be alleviated by using workarounds such as preset templates embedded in the handsets of the recipients that are filled by the SMS. The Sahana solution does not include this feature at this time, but it can accommodate such improvements.

The short-term solution for journalists is to combine the SMS with another technology, using the concept of “complementary redundancy” developed during the LIRNEasia study. This means that the simplicity of the group SMS transmission is used to alert the recipient; the greater detail is provided through a website, which should be accessible from any newsroom (subject, of course, to it being designed to accommodate a large number of simultaneous hits, but this is not hard).

The addressable satellite radio solution field-tested for the first time in Sri Lanka by LIRNEasia can also serve to achieve complementary redundancy. However, the fixed costs of this solution are such that its deployment requires government involvement at this time.

In sum, the simple SMS module designed as part of the Sahana suite is a simple, robust solution that can be deployed in weeks. It reaches the ordinary mobile phone that has today achieved the status of organic extension of the sense of hearing. Like vision-impaired people (like me) always carry their glasses, many of us always carry our mobiles.

Cell broadcasting

In discussing the question of public warning at a meeting of telecom operators and senior disaster managers convened by the Ministry of Disaster Management and Human Rights in 2007, I challenged the operators to solve two problems. If they were solved, the rest would be easy. The two problems were:

  • How to reach a passenger on a moving train (like the hundreds who perished unwarned in Pereliya on 26th December 2004); and
  • How to reach visitors at the rest stop on the beach at Yala (where dozens perished from the tsunami as well).

The reason I considered these the hardest problems was because they are media-poor locations. Today, almost all vehicles on the road are equipped with radios, though trains are not. The country is pretty much blanketed with mobile signal coverage now, though I then thought the signal had not reached the Yala location (now marked by what is one of the more elegant of the tsunami memorials).

The answer is cell broadcasting, a capability of mobile networks that uses frequencies other than those used for calls (such as those used to transmit base-station-location information) to broadcast or transmit in a point-to-multi-point mode. This mode is inherently immune to congestion. Radio and TV stations do not go down in the aftermath of a bombing when everyone turns on their sets; neither does cell broadcasting.

Cell broadcasting requires no addresses (numbers) and no switching. It simply broadcasts a short message to all the appropriately tuned handsets in the range of specific base stations (those of the customers of that network plus any roaming customers).

Its blanketing of the coverage area of a base station is advantageous because hazards affect specific geographic areas. With cell broadcasting all the mobiles in the network in that area at that time are reached, not just those who have pre-registered as in the Sahana SMS module.

The ability to transmit warning by base-station area also allows the tailoring of messages, avoiding for example the alerting of the residents of Kandy to an impending tsunami or allowing the ordering of evacuation in one district while asking residents in another district to remain alert.

It is one-way and can carry only a limited number of characters. In its current form it does not emit a ringing tone or a vibration, leaving open the possibility of the warning reaching the phone but not the person (who fails to take it out of the pocket).

In the aftermath of the tsunami, Sri Lankan software engineers (MicroImage) demonstrated that this last problem can be solved by downloading applets into Java or Symbion-based phones that generate unique audible alerts. They have also demonstrated the display of warnings in all three official languages.

Reach of the mobile

We can expect first responders and journalists to carry mobiles. Can we expect the same of the general public, one may ask? Is the transmission of public warning via mobile some kind of conspiracy to save the rich while leaving the poor to drown?

Today, Sri Lanka’s mobile penetration is above the half-way point. There are around 11 million active SIMS for a population of 20 million. What this means is that mobile penetration now is higher than radio penetration was at the time of the 1978 cyclone that wreaked devastation (but not too many deaths because of effective radio-based warning and good first-responder action) on the East Coast.

According to a representative survey of bottom-of-the-pyramid users conducted by AC Nielsen in 2006 on behalf of LIRNEasia, 22 per cent of the lowest-income households in Sri Lanka at that time had at least one mobile phone. Given the massive growth since then, we can safely assume that around 70 per cent of these households now have either a GSM mobile phone or a CDMA phone provided by the so-called fixed operators.

I say “so-called” because the CDMA phones are actually mobiles dressed up to look like fixed phones. They can be used while moving. With relatively simple tweaks it should be possible to enable these phones to receive cell broadcasts and these networks to transmit them.

The answer to the hardest problems that I set last September is cell broadcasting. It is highly unlikely for there to be a railway carriage that does not have at least a few mobile users. Except in stretches of the Main Line to Vavuniya and its spurs and the Badulla line, signal coverage is likely to be comprehensive. We can send SMS to the driver and guard (as designated first responders) and we can warn the passengers through cell broadcasting of impending danger.

The beach at Yala, I am told, is today covered by mobile signal. With local and foreign tourists pulling out their phones to take pictures and guides exchanging information about where the animals are, the chances of reaching them through cell broadcasting are not bad, even before we get the applets downloaded and alerts enabled.

LIRNEasia, in cooperation of the authorities in the Maldives, is this month starting a scoping study on implementing cell broadcasting in that country. Maldives has almost reached saturation point in mobiles, with as many active SIMs as its population and all major islands and atolls covered.

Once the lessons of this study are available, we hope the authorities in Sri Lanka will also implement these two solutions for the benefit of the people of Sri Lanka and in memory of the needlessly dead.

Myanmar, which has limited the use of mobiles to the Generals and their privileged progeny, will be a harder problem; but not one that cannot be solved.