Ercim news

I wanted to blog about this for a while already. I’ve been waiting eagerly to read the latest issue of ERCIM news, which had for special theme “The Sensor Web“. Usually they have interesting articles on many cool high-tech topics, but I have to admit that this issue was quite disappointing for me.

In particular, they talk about the Sensor Web, which has nothing to do with our WWW. There is not single mention of URI or HTTP or REST in the whole issue, but there are many articles that talk about the same approaches. I find this quite disappointing, as neglecting the Web as a prime choice for transmitting data makes me sad.

Even though there are many mentions to project such as OGC Sensor Web Enablement (SWE), it just scratches the surface of what the WOT is really about. SWE is cool indeed, but to me it’s just a GIS/sensors oriented variation of web services. They have many “proprietary” (I always use proprietary for everything that is not widely accepted by at least 20% of Web users) language to describe sensor and their readings, but then why not use standard web services to do it? Or EEML for that matter which is much better, flexible, and extensible? Don’t get me wrong, SWE’s languages are very good for static cases and rigid deployments, I doubt it’s really better than RPC-based good ol’ web services (just that it has lots of things hard-wired to help you if you deal with sensor & geo data).

All of the approaches in the sensor web issue, still do not sound very scalable to me (this is just an assumption, given that I don’t see many people using them). That makes me mad, so many people trying to push their standard, why solution A is better than B, etc. The result is that no one really makes it, and all these standards just end up marginalized. That’s exactly what we want to avoid. The Web of Things is build on that very Web you’re reading this, the Web that’s being used by millions of people daily? Why people think they can just come up with something totally different, hoping that people will start using their better-than-other standard? We propose to reuse as much as possible of what’s out there, and introduce new things just Web protocols won’t suffice (but I don’t don’t really see what can’t be done with web technologies at this point). Let’s hope in the near future, when people will talk about worldwide global network of sensors, they will mention at least the Web as an excellent candidate on which to build it.

You may also like...

  • “..to me it’s just a GIS/sensors oriented variation of web services.”

    SWE and SensorML did not come from the GIS community. SensorML came initially from a NASA project for sensor descriptions. The SWE approach is very general based on the “Observations and Measurements” spec. O&M provides the metrology basis for SWE and SensorML. SensorML was refined by working with the IEEE 1451 – SensorML is more general.

    EEML is interesting in particular considering that OGC has worked with IFCs. CityGML is what you might compare to IFCs except that it is more recent development than IFCs. GML is an XML vocabulary.

    SWE is web services, with KVP, SOAP and REST variants. SensorML can be used with SWE web services and non web domains. There is a mobile o/s api for SensorML.