Michael Voong HCI Researcher @ Birmingham University

Mobile Personal Web Servers

I’ve been following part of Nokia’s research that’s being done in their Helsinki headquarters and MIT collaborative efforts. One interesting area of research is the development of the concept of running web servers on our mobile phones. This may at first strike as being a waste of time - “web servers aren’t designed for mobile devices”, you say. Yes that’s true, but then again, SMS was only designed to be used by engineers troubleshooting cellular networks. After ignoring the relatively poor bandwidth and connectivity offered by cellular networks in the UK and reading Nokia’s presentations and papers on mobile web servers, or MobiSites, I began pondering over the limits of the technology. Let’s explore this a bit more slowly.

First, how does the technology work? Anyone who’s been behind a NAT firewall (e.g. the school’s) would know that you cant sit behind it and expect computers on the internet to be able to access your server. Computers outside of the firewall cannot see the computers within it, because of this thing called NAT. The advantages are clear: we can share one IP with many computers, we keep IPV4 happy by sharing the IP addresses, and we increase security of the network by shielding it better against external attacks - the firewall alone takes most of the brunt. There’s now the obvious problem of how MobiSites are accessible by arbitrary terminals on the internet. Nokia have solved the problem by using a gateway which the mobile devices ping regularly to see if any connections need to be created. Basically, the connection happens the other way - the mobile phone connects to the gateway, which acts as a transitionary connector between the client and server.

The advantages of this architecture? Well, what are the problems with MobiSites? There are a few key issues: processing speed, bandwidth and connectivity (percentage of time a service is accessible). Let’s ignore the first one, assuming that by the time this research is pushed into the real world, if ever, the processing speed of mobile phones will increase enough. The second and third: bandwidth and connectivity can be solved by caching on the gateway. Of course, connectivity will sacrifice the liveliness of the data; that is, the data may become stale and old during periods of connectivity loss. Photos can be cached, and the gateway can be informed of new data, and everyone’s happy.

But are they? What makes this system any different from one that basically uploads media to a web server every time an update is made? This is a difficult problem and to solve it we must ponder up comparisons between traditional web server paradigms and MobiSites, and find a real need.

Let’s think of it in terms of “what is the most useful thing about what mobile phones can do that the masses would want to interact with?”. Well, the obvious thing is that you use your phone everywhere you go. You walk about the Earth, doing your everyday things - working, playing, socialising, exploring, relaxing, shopping, etc. Let’s think about the places you do these things, because it’s quite clear that the idea of places is the most powerful thing we should be harnessing. Studies have been carried out which conclude by saying that we must research context-aware systems which take into account place rather than space. A place is defined by both a space and what you do in that space; as well as people that share that space. So, using explorative thought, let’s suggest a few places that a person may be in any given day, and suggest the corresponding space:

• On holiday in Hong Kong; in Sai Kung, New Territories, Hong Kong
• In my room, procrastinating with no intent to stop; in a room at 443 Fictional Road, Birmingham, B15 3TO
• Walking around the City Centre, looking for presents for my family; in the Bull Ring, Birmingham, UK

It becomes obvious that nobody cares about a person’s location in absolute, spatial terms. So what, you can pinpoint a person - what’s the use in that? A spy might find that information useful, but it won’t intrigue a person for very long; and that’s ignoring the privacy issues. So now we’ve identified the importance of place, what possibilities can we dream up for this MobiSite idea? Here are some questions I would like to ask about people on the Earth… do any of them interest you at all? Would MobiSites help me answer the questions easier than using other means?

1. I haven’t seen my friend for a while. I used to live near to this friend, so I always knew how they were doing. Now, they’ve moved to another Country, so it’s more difficult. I want a spontaneous view of how they are doing, and their general emotional state, without having to ask them explicitly.
2. What are my friends doing right now? How can I receive an overview without asking them one at a time? I want to do something fun spontaneously, like organise a last minute party.
3. I’m in a weird mood today… I want to make a new friend. Who else is feeling the same as me? How can I find them? I don’t want to spend ages introducing myself to them, but I want to share my thoughts with them.
4. I went on holiday with some friends. How do we share our thoughts without resorting to setting up a central repository, eg. website? I want to subscribe to notifications of updates from my friends.
5. How do I advertise myself so that I can find like-minded people?

OK it’s getting late so I’m gonna end the post there and think of follow-ups to those thoughts tomorrow! Cheers for reading!

Once you have flat-rate broadband on your mobile I think it become a moot point where the webserver software lives.

However, the question “what is the most useful thing about what mobile phones can do that the masses would want to interact with?” crackles with importance. One answer is twitter.com “What are you doing?”. My answer, of course, is bodtracker which is back in development very soon. I might nick your scenarios if you don’t mind!