towards survival instincts in computing systems

I have recently talked about developing survival instincts in computing systems. This opens up an interesting paradigm for designing autonomous systems for applications that require them to be on earth, underwater and in space. The conditions for operation of such systems are often harsh, unpredictable and it seems most natural to look for analogies to envisage the ways of their design in the nature, in animals and humans, particularly looking at the nervous systems. Another important pathway to such systems would be to look how energy affects their behaviour and how power levels activate various layers of instinct mechanisms …

These were the ideas that I discussed in my keynote talk at NoCArc’12 in Vancouver  (http://www.unikore.it/nocarc/index.html).

Here are the slides http://www.unikore.it/nocarc/slides/yakovlev.pdfand and video http://www.youtube.com/watch?v=lgcugX44EIg&feature=youtu.befrom

 

SAVVIE: A follow-on to Holistic project

Staying alive in variable, intermittent, low-power environments (SAVVIE)

EPSRC Joint Research Grant: EP/K012908/1 & EP/K011979/1
Institutions: University of Bristol and Newcastle University
Start Date: 1 December 2012Today’s low-power electronic systems are designed to handle a high variability in the power demand, for example during transmissions from miniature wireless sensors. However these systems cannot cope with a highly variable power supply. If they are powered by an ambient energy harvester in an environment where the available power is low and sporadic, the system dies once the energy storage becomes depleted or damaged, with start-up being impossible if the power is not increased to a higher steady level.

This project researches how to design robust and reliable electronics for situations where there is a variable, unreliable source of energy. A number of situations, or states, have been defined, according to the level of depletion of on-board energy storage, and how variable the power supply is. In the most challenging states, for example where the input power is sporadic and spread over a wide range from nW to mW, modern electronics fails. We call this the “survival zone” and are investigating a combination of techniques from the areas of power electronics and asynchronous microelectronic design to allow devices to operate in this zone. Techniques include control circuits that are able to ride through variable voltages, the detection of states, and reconfigurable hardware resources and control algorithms to suit sporadic and sub-microwatt input power. The chief aim of this project is to produce survival zone design methods for the microelectronic design community.

See the project web site:

http://www.bristol.ac.uk/engineering/research/em/research/savvie.html

 

University of Bristol logoNewcastle University logo

Energy-proportional sensing

What is energy-proportional sensing?

Sensing (S) is essentially a process of measurement of a physical quantity and presenting its value in a form that can be used for electronic or human data manipulation. It is a (sequential) composition of transduction (T) and conversion (C), i.e. S = T;C.

Suppose we have a way of transduction of a sensed quantity, i.e. turning it into an electrical parameter, such as voltage or current. One way of transduction could be to turn the sensed parameter into energy. The energy can then be turned into a computation whose final results could represent the energy used in this computation, which in its turn could represent the original parameter. To make such a sensor we need two aspects of proportionality (ideally, in linear relationship). One is that the original parameter is turned into the amount of energy in a proportional way, and the other is that the obtained amount of energy is turned into an information representation also in a proportional way.

Our proposed voltage sensor (see my previous posting about our patent application on voltage sensor) is based exactly on this principle. The input voltage is converted into a electric charge (energy) in a sampled capacitor, and then this charge (energy) is converted into a binary code produced by the electronic circuit which is fed by the energy of the charge.

Long-living computer systems …

Biological systems typically have two types of operation, regular and bursty, and manage to organise their operation in an energy efficient and robust way, which also supports natural tendency for survival. Regular activities take place all the time, and are meant to serve the needs of the overall system and are determined by the overall structure and dynamics of the system. Bursty activities are typically not those that are constantly triggered by normal periodic cycles of the system, but rather they are triggered by or in accordance with the needs to react to the demands of the environment. Why not to build a computer system in a similar fashion, such that a constantly active part has to be relatively slow and all the fast processing has to be done in specialised units, whose activation is bursty?

For more information visit the URL and find my recent technical memo on this …
http://async.org.uk/tech-memos/NCL-EECE-MSD-MEMO-2012-005.pdf