Book Project: Real World Awareness
From IRF Wiki
Real World Awareness
- The questions raised at the IR Forum are discussed in the context of the patterns.
- Further ideas and research are added from the research interviews.
I
Ideas from the Proceedings
Hub vs. Edge:
- Real-World Awareness (RWA) raises important issues concerning how and where to process “torrents of data” - at the edge, close to where sensor data is actually generated, or at the hub, where it can be analyzed and interpreted en masse? Each has its advantages and disadvantages.
- Intel’s Rattner points out that “ ...we are getting more and more computing capability per cubic foot. We are getting tremendous advances in storage capacity. ... the analysis techniques, particularly the ones involving machine learning and statistical computing, are going to be very important to going forward. ... the wide deployment of sensor technology is going to result in a torrent of data and that is something we are very concerned about. Especially if the idea is to sort of aggregate all of that data and send it to ... the traditional backend. We think there is a real question of bringing a lot more intelligence, a lot more computing to the edge as supposed to back in the datacenter.”
- Lutz raised the possibility of analyzing and processing these torrents of data in the network itself, closer to the edge and before it hits the central hub. He warned, however, that there may be legal reasons to at least collect all of the sensor data in some central repository.
- An audience member pointed out that in some cases, aggregation and condensation of data can be performed on the spot, very close to the sensors themselves. One example: readings from 7 sensors in a car may indicate that it is aquaplaning on a highway, but only that single, high-level fact - that aquaplaning is occuring right now - needs to be shared with the car’s driver and with those of nearby cars.
- Another audience member said that physicists have tackled this problem in their high-energy experiments, where floods of data are filtered and greatly reduced in number within milliseconds of their initial collection.
- Dynamic, just-in-time filtering and interpretive models that can be quickly adapted to meet new conditions appears to be one solution to the problem and a way to add useful semantics to raw sensor data out at the edge.
Scope of RWA implementations:
- Rattner: “... there are a lot of issues in terms of the ecosystem and terms of making these technologies scale well. ... we have work to do on the standard side and of course the security issues ... are really critical here and represent ... some of the biggest challenges to our ability. Any time you try to automate something you know people are going to resist that until you demonstrate solutions to their critical concern and it'll be no different in this area.”
- RWA holds tremendous promise, everyone agrees, but there are some difficult standards issues to be resolved in order for this promise to be fully realized. Different RWA products and schemes around the world currently use different formats for representing data and this lack of standardization will cause severe problems for users trying to integrate data in centralized ERP and supply-chain systems.
- Alex. Schill from Dresden University of Technology pointed out that low-level RFID sensor data is pretty much useless without some kind of context wrapped around it. And so, there is the need for “a global and distributed context service,” which he said presents a stiff technical challenge.
- Security and privacy are vital issues for the success of RFID. The public is concerned that too much revealing information will be collected about them and their activities.
Agile Development:
- “ ... we need to think things from end-to-end ... I do not believe anymore you can know ex-ante what end-to-end means in the year 2006 and 2007 with the same level of confidence you could in the year 1996. ... That's where we are navigating between: really great understanding of end-to-end versus this new technology, which is, ‘what does flexibility add, what do platform flexibility and adaptation need to mean in the next three to ten years?’ ... the future of this is not in figuring out ... what exactly the best sensors are, but is more figuring out what the applications are and the flexibilities.”
Contextual Consciousness - User role - IT for growth and development:
- “... the work of Charlotta Perez, who talks about the need of government ... to act as brokers between the world of ICTs [???] and citizens. ... present certain aspects of RFIDs ... that will appeal to citizens, that we will not repeat the mistake in terms of presentation that we made with genetically modified food or nuclear power. ... we need to explain to citizens more and more that ICTs do not just produce economic benefit, but also social benefits, that they help us to have better health care and more security, more protection, that it is a way of responding to the values we live in.
Agile development/security:
- “ ... the fragmentation of systems. We decompose our systems in small fragments, which are interlinked with networks and so on, each component in that system has its own intelligence, and this will also lead to a fragmentation of the value chain. So the value chain can also be fragmented and so we will have completely new players with completely new business models, which are entering the market in the future. ... fragmentation can also lead to fragility, so we have to cope with some important technology challenges, for example architecture, control of highly distributed systems, ... ensuring dependability and security, especially security for that critical infrastructure."
