I’m on my way back from a particularly fruitful and productive mission to Mumbai. As noted in my earlier blog, the purpose of the mission was to explore possibilities for partnership and collaboration vis-a-vis “upgrading” Mumbai city’s disaster early warning/response system. We chose to focus first on the Monsoons (which necessarily includes an important public health component).
Thanks to our fellow HHI colleague, the legendary Dr. Satchit Balsari, we met with all the key stakeholders in a whirlwind tour that included heavy rains, seemingly suicidal drivers, the amazing Ganesh festival and countless hours of near infinite traffic. All a very low price to pay for the energy and proactive engagement that emerged in our meetings, which ranged from high level political officials to leading professors based in Mumbai. Our presentations on crisis mapping platforms were also very well received, with comments including: “That’s exactly what we need.” Equally importantly, we saw numerous slums (picture below) along with the depressing conditions that slum dwellers have to live in. Naturally, they are the most vulnerable populations in Mumbai city, and our project will only be successful if it makes a difference in their lives.
Below is a picture of the control room headquarters for Mumbai city’s disaster early warning and response center. Note the red phones, each with a direct link to a government agency. It was interesting to note that apart from Autocad, the control room was not making any use of mapping software; all maps appearing in hardcopy. The center recently carried out a major vulnerability analysis of the entire city, which now comprises over fifty pages of very rich, albeit dense, structural data. This data has not been mapped.
Our proposal was rather simple: develop a web-based interface using Google Maps that allows for easy mapping of structural and dynamic (event-data). In addition to manual mapping, allow live, automated (meteorological) feeds from the control room’s computers, to be visualized within Google Maps in real time. Integrate SMS broadcasting directly within the Google Maps interface; thereby encouraging us to think of dynamic maps as communication tools in addition to tools for situational awareness. This follows Ushahidi’s approach of crowdsourcing crisis information.
The importance of two-way SMS broadcasting for this project cannot be overemphasized. Some 75% of Mumbai’s residents own a mobile phone (note that 55% of Mumbai is composed of slums). By some measures, Mumbai is the fourth largest city in the world, with a population of some 16 million. With this size comes some obvious challenges. I had pitched FrontlineSMS as a potential solution during my presentations, but I don’t know how scalable the software is. Can the platform handle millions of text messages in a matter of days? Ken Banks is kindly looking into this for us (thanks again, Ken). Another challenge is whether the bandwidth of Mumbai city will be able to handle such high traffic.
We will be following up on these questions and conversations shortly, and will be scheduling another mission to Mumbai in the next few months to formalize our partnerships and finalize the operational framework. As always, I welcome any and all (constructive) feedback.
WRT FrontlineSMS, the message software isn’t so much the limiting factor as the way it allows or forces you to connect to mobile network. Your limitation is more a function of either how many GSM modems (lines) you have attached to a computer / server for a mobile-hub to SMS gateway, or it will depend upon the service provider gateway (SMSC). An SMSC should be able to handle lots of messages in a matter of minutes, but you’ll need to check with the service provider and perhaps establish a special contract to ensure they don’t have a time based cap. With an SMSC not modem required. Just a good internet connection to the service providers SMSC or a commercial gateway like Clickatel. I’m not positive, but it looks as if FrontlineSMS is designed to have one or more modems / wireless phones attached — no SMSC option. In this case, the software is not even scalable into the hundreds of messages for early warning applications where time really matters. I suppose it’d be fine for very slow onset events, or if you were just wanting a bunch of emergency managers to message each other quickly.
As a general rule of thumb, you should estimate that each modem requires 12 seconds to process a message. I’ve seen some do it in 5-6 seconds with good regularity, but more is generally required. This then presents you with some simple math. Reliably you can only get 5 messages out with a single GSM modem / phone in a minute. 300 in an hour. 7200 a day. Yes, you may have performance that doubles this number, but you get my point that you aren’t going to reach many people with even a bank of a dozen modems in a very short period of time. Clearly, if you needed to reach mass numbers of people in say 10 minutes or less, you need a massive modem bank. If you wanted to create such a bank, you’d probably write your own software, but commercial options are first and foremost NowSMS. ActiveSMS is also a good little program, but it maxes out at 8 transports (GSM modems / phones). Both of these of course would require some programming to send and retrieve messages with a GUI, but at least you wouldn’t have to write AT commands to the modem.
I’m not convinced of SMS for mass public early warning applications. I personally think it should be used to communicate with key invidiuals and offices, as it is available — so why not. But the mobile infrastructure itself is highly perishable in an event. Some service providers in various Asia countries have reportedly had to to turn off or cycle data and SMS services during normal telecomm. ‘rush hours’. (Who knows what happens during a disaster or emergency. ) This effectively means a fored delay in messaging. As you pointed out the local infrastructure may not be able to handle messaging. Of course the real issue is that SMS is store and forward like e-mail. So if the queue is long, your message is delayed. For short bursts to key individuals, SMS makes sense, but for mass public warning its utility is currently dubious unless you have explored and tested scenarios very well with the local / national service provider. As far as I know Finland is the only country doing mass public early warning via mobile, but I don’t think they are using SMS actually — rather some backend true broadcast channel — not sure — been awhile since I read up on it.
It is also important to ensure that the warning authority (often different than the disaster response and emergency management entity) is on board with you for messaging. Aside from some technical challenges to the feasability of mass public messaging, I’d be concerned about the message itself. A very well thought out and studied standard operating procedure would need to be developed. People react to information differently, and studies show that people actually do comprehend and use more detailed warning information. Short bursts limited to 140 characters or less could cause more harm than good by ‘miss messaging’. This is of course unless the message format is well developed, public education undertaken, and some research done to see how people interpret certain messages.
Anyway, good work. Don’t mean to be a downer. Just wanted to raise the issue.
“… upgrading” Mumbai city’s disaster early warning/response system… focus first on the Monsoons… develop a web-based interface using Google Maps that allows for easy mapping of structural and dynamic (event-data). In addition to manual mapping, allow live, automated (meteorological) feeds from the control room’s computers, to be visualized within Google Maps in real time. Integrate SMS broadcasting directly within the Google Maps interface… tools for situational awareness.”
Mumbai, inhabiting the country’s top-notch business houses and elite work-force, pathetically lacks planning and proper implementation, when it comes to urban infrastructure.
You must’ve noticed the same on your visit: You take a right… then a left… a left again… and you reach the same place you started from – a concrete puzzle!
The all-knowing elite flat-owner, having accessing to an OKAY Internet connection would find the Google Maps interface very informative. But if one really wants to get to the root of crowd sourcing crisis information; it’s the people who’re hit the most in times of crisis. For example: hutment dwellers and residential colonies in low-lying areas, coastal regions of the seven islands.
I was present at the launch of mumbaiVOICES (http://www.hhi.harvard.edu/index.php?option=com_content&task=view&id=129), a web-based application that facilitates a citizen driven analysis of urban disaster response. The current state of affairs at the website is SAD, with the entire Discussion Forum, full of spam. Check it out – http://www.mumbaivoices.com/index.php?option=com_joomlaboard&Itemid=47&func=showcat&catid=2 . I hope you would bring this issue, to the notice of Mr. Balsari.
All in all, I think an SMS driven sourcing tool for situational awareness, would do good (comparatively), provided it benefits the people in crisis and not just as a source of data collection.
Many thanks for your thoughts and constructive feedback.
“The all-knowing elite flat-owner, having accessing to an OKAY Internet connection would find the Google Maps interface very informative. But if one really wants to get to the root of crowd sourcing crisis information; it’s the people who’re hit the most in times of crisis. For example: hutment dwellers and residential colonies in low-lying areas, coastal regions of the seven islands.”
Couldn’t agree more, hence this reference in the blog you’re responding to:
“Equally importantly, we saw numerous slums along with the depressing conditions that slum dwellers have to live in. Naturally, they are the most vulnerable populations in Mumbai city, and our project will only be successful if it makes a difference in their lives.”
On MumbaiVoices, yes, the chat functionality was actually shut down a while back but has still been hacked into. The plan is to take it down altogether this week. But lets not have this distract from the important added value provided by MumbaiVoices: the 150 testimonies collected (from survivors, witnesses etc) which cover an incredibly rich and at times surprising set of lessons learned. These are now being integrated in Mumbai’s major disaster drill scheduled for November. That, in my opinion, was the main added values of the project, not the chat room.
“All in all, I think an SMS driven sourcing tool for situational awareness, would do good (comparatively), provided it benefits the people in crisis and not just as a source of data collection.”
Couldn’t agree more, see my same point above.
Many thanks once more for your input and feedback, Allan, very much appreciated.
Pingback: Mobile phone innovation may not happen how you think it does. : crisscrossed blog
Pingback: Top 5 iRevolution Posts of 2008 « iRevolution
Very Interesting post! Thank you for such interesting resource!
PS: Sorry for my bad english, I’v just started to learn this language 😉
Your, Raiul Baztepo
Thanks for your kind words, Raiul!
The test of good writing is, like good vino, to have it get better with age.
Your article here, which I read today after a couple of months, still resonates so very much with the work I do in post-war Sri Lanka.
It’s so refreshing to read intelligent thinking that’s well written to boot. Thank you.
Hello there just happened upon your blog from
Google after I entered in, “SMS and Web 2.0 for Mumbai Early Warning/Response Project | iRevolution” or something similar (can’t quite remember exactly). Anyways, I’m pleased
I found it because your content is exactly what I’m searching for (writing a college paper) and I hope you don’t mind if I collect
some material from here and I will of course credit you as the source.