WG211/M8Dobson

Programming for adaptive sensor networks: back to the future

Simon Dobson

Programming for sensor networks sometimes seems to combine the entire history of computer science into one single challenge: fitting a modern, advanced, robust, adaptive distributed system onto platforms with the computational power of a mid-1980's home computer. If we're to make such networks behave correctly in the face of changing environments and tasks, we need radically to re-think how we develop code for them. Treating the network as a whole, and generating as much of the code as possible according to a higher-level description of its behaviour, seems like the only way forward.