Globecom 2012 – day 3 coverage: Attended the Industry forum on IT Industry Transformation – Clouds, Security, Mobility and Computing. The keynote speaker was Flavio Bonomi, Head of Advanced Architecture and Research at Cisco Systems. The talk was very much focussed on the latest buzzword in the industry & research community – “Fog Computing”. As opposed to Cloud computing, where all computing is “centralized”, fog computing relates to the concept of using the very capable computing and storing devices in your local neighborhood, at the edge of the network, to share resources locally to jointly compute and store information. This would take a lot of the load of the network not having to pass a lot of raw data to the cloud. The prerequisite for this is a lot of local “D2D” – network formation to share the data and computational tasks – in fact this could potentially be the “killer driver” for D2D communication.
The question is how this fog is formed. Most appealing is of course the “ad-hoc” formation of resources, making use of everything that is in your neighborhood to give you a hand to help out with your computational task. Looking back at history, adhoc networking and computation has been researched for more than 30 years – unfortunately, little progress has been achieved, at least if you measure it in commercial success. There is the complexity issue – the problems in ad-hoc systems generally lend themselves to be solved by what we are increasing better at, i.e. throwing more bandwidth ant computational power at the problem. There is also the fundamental issue of “selfish” sharing – why would I sacrifice the precious power (bandwidth etc) in my device to forward your packet, do your computation etc? Research results boil down to that there will always be devices that have resources that everyone likes to share, and others (“leeches”) that have little to share. This actually works, if resources are (perceived to be) for free – Bit-torrents are good examples of this.
Interesting research issues ahead… and some issue in understanding business models as well. Who would drive this development… operators, terminal vendors, telecom infrastructure vendors..? Or is it something that everyone wants, but no one will take the first step ?