devlog

Wireless network mapping is all the rage again. I’ve spent a fair amount of time researching and building prototypes in this space over the last couple of years. After 6 months of stasis, WirelessLondon recently unearthed a small budget surplus, which would pay for 7 days of my time among other things. We’re trying to decide what would be most useful for “the community” for us to spend a 2-week sprint on. Here’s my attempt to break down the problem space, and point to similar and related efforts that I know of.

Community network node maps (and feeds)

WirelessLondon started out attempting to replace the consume.net node map, which had no basemap - unless you knew the area well already, you couldn’t tell where you were. This was all before Google Maps, and in the UK, there’s still no open access to state-collected geographic information, so we needed to make our own maps from scratch. We used Mapserver with a free but low-resolution Landsat satellite basemap, and started collaborating with the OpenStreetmap project, an effort to build free-of-copyright maps by collecting GPS traces and collectively drawing street shapes over them and annotating them.

I’d been talking about all this for a long time with Earle from the Open Guides, the original open source “spatial wiki” that started in London. People describe the things they like about their city, annotate them with spatial information - a postal code, and address, co-ordinates. I realised that if the node database knew where each node was on the map, and the Guide knew what was nearby, WirelessLondon could use a “captive portal” to show users collaboratively-mapped local information when they connected to the community network.
This starts to get a long way from the original network map; but that’s the key to the whole thing. The most recent attempt at a simple, generic network mapping app I saw was at the last National Summit on Community Wireless - CUWin’s WNMap. It’s Google Maps based, on sourceforge, and works. But to me, it’s just more red dots on a map - I want to join the dots that we can’t see, reach into the map and haul the information out. There’s so much more that can be done once one connects local wireless networks with spatial information and open source geospatial software.

• integration with network management software
  This is what wifidog have been doing for some time. You know a node’s location in order to be able to deliver a custom portal page to it, tell who’s connected to it, monitor its status and visit physically for updates if necessary. This is the hallmark of a very “managed” community network (where all the nodes are maintained and monitored by a dedicated group) - the bristolwireless network is similar - rather than London’s, which was ever an ad-hoc collection of nodes that individuals decided to add to a shared map. The FreiFunk group in Berlin have an even more sophisticated take on this, using the bbbike open dataset to map routes on their dynamic mesh network in near to real time.
• integration with local information “captive portal”. The original WirelessLondon software used NoCatSplash to poll a central server for an updated page, generated for each node, based on RSS feeds collected from the Open Guide based on a node’s location. After I met the Ile Sans Fil group and converted to WifiDog, the node would visit an (internet-based) central server every time a user connected, to get local updates. The ideal situation is one in which the node has ‘local knowledge’ or can collect it from a from a peer over the local network, and doesn’t have to visit the internet to find out what’s nearby it in space.

The “Pulse Points” project here in Boston is along these lines. As I understand it, the Boston Globe wants to make access to its historic archives available locally, where they are vivid and useful content; but reserve the right to charge for archive access over the main internet. This is a free-at-the-point-of-access project, but has serious proprietary backing. But as open hardware projects like “hive networks” appear, implementing local knowledge with local storage on community-maintained wireless networks becomes more of a potential reality. Hardware and software wise, we’re golden; but whence comes information?

Who’s making spatial data feeds?

Last year, more or less no-one; WL put some energy into the Open Guides project just to have a place to start. Lack of access to state-collected geographic data, and restrictive database copyright law, also held us back from web-scraping and geocoding our own local information resources - everything had to be built from scratch. Tourist sites like wikitravel aren’t really so useful; If i’ve been living near nelson’s column all my life, i don’t want to be reminded that it’s there, every day, each time i connect to my local community network. Now, services like platial are starting to make more meaningful and personal spatial annotations available as GeoRSS feeds; a lot of other sites are getting on the bandwagon.
Event syndication in a spatial context is a big missing piece that’s still not quite there, that i can see, and something a lot of people are looking towards wistfully. The situation improves now as services like upcoming are gobbled up by Yahoo! who in turn have some interest in producing as much data richness as possible for their own services. I don’t want dots-on-a-timeline any more than I want dots-on-a-map; but I want my community information portal to be full of local history and near-future civic activity.

How open is open by default?

Network maps don’t just have to map what people say is open. In a way, a community network map is not for the casual user; it’s to bootstrap a different kind of connectedness, of belonging in the users and maintainers of the network. Of more use to someone who comes along saying “i’ll be in this part of London / St.Paul / wherever today, where can i get online?” is a stumbled map, created by a collaborative process of people uploading their logs of where they saw open access points while walking or driving around town. Wigle is one of the best known netstumbling sites; they get uploads from users, infer node locations, discard the stumbled data and re-sell the node location + essid + MAC information to companies providing wifi geolocation services.

WifiMaps is another wifi stumbling effort driven by open standards / open source GIS backend. It was groundbreaking for its time and is showing its age a bit now. Its maintainer Drew is keen to share the love and share the data; The borg at the free map system is interested in taking it on. Outside the US, the problem was alwayys getting the rights to re-use a basemap… but then Google Maps came along.

The Google Maps Trap

I’ve heard both Benoit and Mike from ISF complain that as soon as wifidog implemented a Google Maps-based node map, any conversation in their group about open source geospatial tools instantly ceased; GMaps was “good enough”. And it’s certainly easy to use. These were two big missing pieces with which we struggled heavily at WirelessLondon. This is changing now, and getting better all the time; OpenStreetmap is really starting to fill out, in areas where there are enthusiastic data gardeners. The OpenLayers project provides an excellent, very low overhead Google Maps quality-and-feel user interface for open standards based web mapping services. GeoNames provide an excellent and growing open gazetteer and geocoding service for searching geographic names - at least getting to the right area on a map. We don’t have to stay stuck in the Google Maps trap.

Why share node data / metadata?

The conversation about sharing and syndicating node databases between community wireless networks cycles round now and again. People talk about an RSS-based standard for transferring nodedb data and updates around so they can be compiled into a collected map. Someone points at the XML namespace/vocabulary that WL uses. Then people shrug and say, “well why bother - why should someone in St. Paul care about open wireless nodes in London, or vice versa?” I don’t have a good answer to that. I believe that data, like networks, should be open if there’s no really good reason why not. I believe that standards where they exist, whether ad-hoc or highly formalised, should be adhered to to save people wasting time and let them use that saved time to do unpredictable, enjoyable things. I believe anyone should be able to - should be encouraged to - run their own local information service. The very simple wifidog client protocol is useful and cool in this respect.

So what happens next?

The WirelessLondon project went through too many visions in the space of a year. By the end of its lifecycle it was turning into a visionary but completely unrealistic collaborative scheduling and proto print-on-demand service. It had crazy dependencies, and I was always more enthusiastic about building machine-level tools - RSS/RDF “stuff associators” collecting feeds from different sites - different ways of querying spatial and temporal information - fuzzy inferences about what similar things in space and time might be identical. My heart was never really in user-interface work, and it showed. After a good timeout from the project, I want to go “back to basics” and rescue something really simple from it. I know that WifiDog would find an RSS aggregator that collected spatial/temporal information and generated “what’s nearby” / “what’s soon” feeds for each node, a very useful thing. I’m interested in working on that, though I know that mapufacture is covering a lot of that same ground now. And I’m also really keen to help more people escape the Google Maps trap.

I’m really interested in getting feedback from people on what they need, what they think is missing, and what open source projects are already covering enough of this that it would be crazy to “reinvent the wheel”. I have a week of time which I want to donate to whatever people need most; this could be a one-off, and could equally be used to seed a future project. If this catches your mind, tell me what you think!