Slashdot is powered by your submissions, so send in your scoop

 


Forgot your password?
Close
typodupeerror
×
Technology

Fruit Flies Making Inroads on Autonomous Computing 73

Posted by michael from the buzz-buzz dept.
Jucius Maximus writes "The configuration of base stations in cell phone networks has always been problematic because you can never predict how many phones will connect to which base station. And sometimes adjacent antennas will use the same frequency leading to dropped calls. Such configuration challenges may have solutions in autonomous computing. An article on C|NET describes how British Telecom is examining the development of fruit flies, hoping that nature has already found the solution to this problem. This technology could also be applied to 'threat-sensing' on computer networks."
This discussion has been archived. No new comments can be posted.

Fruit Flies Making Inroads on Autonomous Computing More

Fruit Flies Making Inroads on Autonomous Computing

Comments Filter:
  • come up with the noise the phone makes when it rings. :)
  • Dead Fruit == Beowoulf cluster

    *Ducks

  • Offtopic (sorta) Humor (Score:3, Funny)

    by dbarron ( 286 ) on Sunday April 14, 2002 @11:35AM (#3338978)
    The solution is to just train fruit flies (instead of carrier pigeon) to carry the messages. Forget the radiation causing cell phones and switches, use Drosophila, nature has the solution.
    Only things is...your message recepient must be in the same room with you...and have a banana in his pocket.

  • huh? (Score:4, Funny)

    by Joe the Lesser ( 533425 ) on Sunday April 14, 2002 @11:41AM (#3338996) Homepage Journal
    'British Telecom is examining the development of fruit flies, hoping that nature has already found the solution to this problem.'

    AT&T has also joined the fray, though it has chosen the hippopotamus as it's bringer of change, and Verizon is leaning toward the studying Brazilians due to their high amounts of energy.
    • You're right to be sarcastic: I'd rather skip the fly-brained tests and jump directly to the hare-brained ones.

      Not to say the whole article can be summarised as "decentralisation and more autonomy, combined with local feed-back". Why do they have to resort to flies?
      Oh wait, isn't this the same company [slashdot.org] that realized a week ago something that was generally known for years?

  • Autonomous Systems (Score:3, Insightful)

    by hs81 ( 62329 ) on Sunday April 14, 2002 @11:43AM (#3339006) Homepage
    This is another interresting idea on making systems that can self-adapt and grow/repair without intervention from engineers.
    This is fine but if we take the 'nature' analogy that this science is following what happens when disease or illness strikes? For example a computer virus or trojan.
    We will be dependent on crucial systems that, at some stage in the future, we will not understand how they work or how to repair them quickly.
    This maybe acceptable for cell phone networks but what would happen if this was an air traffic control system?
    I'm all for following the route in the Cnet article but we need to ensure that we build in processes to control and understand the changes that the system is implementing.
  • if this is threat-sensing technology, then i want an Africanized Bee-owulf...oh, forget it...

  • Captain Obvious! (Score:3, Insightful)

    by PhoenxHwk ( 254106 ) on Sunday April 14, 2002 @11:59AM (#3339064) Homepage
    This article should really win the Captain Obvious award. I mean, really, it's not too hard to come up with the idea of "tell your neighbors which frequencies you're using". I could have saved them a bunch of money if they had asked me instead of some dumb flies! Yeesh.
  • Damn. For a minute there I thought they were implanting transmitters into fruit flies. Where I live, that would certainly solve the problem of distribution; fruit flies are everywhere.

  • *scratches head* (Score:4, Insightful)

    by Akardam ( 186995 ) on Sunday April 14, 2002 @12:11PM (#3339091)
    This... would seem somewhat obvions to anyone who's had to deal with overlapping systems sharing a rather small resource, I'd think. As soon as I read the description of the problem, my first reaction was, "build an auto-negotiating, ad-hoc type system that'd figure it out for itself". As an example, don't networks of SMB clients (with no servers present) already hold "elections" to see who should be the Browse Master?

    On the other hand, the article wasn't to clear on whether BT was using the general idea from the fruit fly, or was using some algorythm derived directly from those cells.

    Let's just hope they don't try and patent it.

    • Let's just hope they don't try and patent it.

      Look at Craig Reynolds' work at www.red3d.com. He developed a similar distributed system for use in computer animation in 1987. Very good for modelling groups of animals such as flocks or herds.

      The general concept at use here - complex, adaptive behaviour resulting from local interactions between agents - is referred to as emergence, and has been a subject of research since the animation work of Reynolds' and the robotic work of Rodney Brooks (from about 1986 on).

      It is good to see big corps showing an interest in it though.

      And yeah, it seems pretty obvious to me too. But some people are married to the top-down, centralized approach I guess...

      • Let's just hope they don't try and patent it.

        But there is plenty of prior art, and they even adminit it: the fruit fly...

        And yeah, it seems pretty obvious to me too. But some people are married to the top-down, centralized approach I guess...

        The naysayers may actually be right... Indeed, stability of such an adaptive system could be an issue. What happens if conditions are such that suddenly the systems decides to oscillate between two meta-stable states, dropping calls at each flip? Could mischievous network users actually deliberately cause such a situation to happen (by gathering enough friens with mobile phones, and driving around the country in certain well-crafted pattern -- remember one of the variables of the system is phones per cell)? Weren't there some problems with similarly adaptive systems in the early ninetys on landline switches in the US, where one switch after the other keeled over like dominos [ncl.ac.uk], all triggered by a trivial malfunction on just one switch?


      • And yeah, it seems pretty obvious to me too. But some people are married to the top-down, centralized approach I guess...

        Sometimes the reason for top-down, centralized management is regulatory - in the US, the FCC probably requires each cell sites' frequencies to be declared on a license. It'd be kinda hard to allow dynamic configuration when the regulatory body requires fixed freqs/cell. Just my $0.02.

  • For the curious, this is how the fly works (Score:4, Informative)

    by DrJay ( 102053 ) on Sunday April 14, 2002 @12:30PM (#3339143) Homepage
    They were a bit vague on what goes on in the fly, so here's a short synopsis:

    The entire back of the fly has the potential to become either a sensory organ or cuticle (the fly's skin). Things get narrowed down by the expression of some sensory-promoting proteins in clusters of cells at specific locations. The process they're intrigued by is how a single cell within this cluster becomes sensory.

    All of these cells begin expressing a signal and a receptor for that signal. When the signal is received, a cell will turn down the expression of both the signal and the sensory promoting genes. Cells that aren't receiving the signal will turn down the expression of the receptor and turn up the expression of the signal. Thus, when a cell isn't seeing a lot of signal, it both reduces its ability to see any more, and increases its signaling to surrounding cells.

    In the end, it all becomes a balancing act between signaling and seeing signaling from neighboring cells. The thought is that the initiation of the process is somewhat stochastic - some cells may turn on the signal earlier than others or start of expressing it at a higher level. The result of the signaling reinforcing itself is that this initial small advantage is greatly amplified, and becomes an all-or-nothing decision.

    I hope that is more coherent than i think it is...

    JT
    • I don't think they will get much out of the drosophila model.

      They should study amoebas instead. The question is how to create an adaptive system that can seek a goal while dealing with the changing environment. The fly development example is not adaptive, or not plastic in development. Reproduction in Amoebas is adaptive and starts with a similar signaling mechanism: Cyclic AMP, CAMP.

      During conditions of nutritional stress, which could affect whole yards of soil, amoebas start releasing CAMP, but they also release a molecule that eats it up. This creates an event/dampener pair that allows for a signal to be propagated as a CAMP gradient...that expands away from the amoeba. These perishing spherical (bearing in mind the complexity of interstitial spaces between dirt grains) signals overlap, interfere additively etc....

      Somehow a chemical consensus is made from the local population, or a center determined, and the amoebas move together. Then they bond together forming a new organism from individuals. After some time, the reproductive stalk/spore forms and bursts out all these new little guys. Then everyone dissassembles and goes back to phagocytizing other unicellular microbes.

  • BT seems to have a thing about insects (Score:3, Funny)

    by flossie ( 135232 ) on Sunday April 14, 2002 @12:30PM (#3339144) Homepage
    Remember how BT used smart ants to solve the travelling salesman problem [nec.com]?

    There seems to be a trend here.

  • Fruit flies like a banana.
    Time flies like an arrow.

    Since frequency is the inverse of time, the solution is obviously with fruit flies!

  • Fly with Implanted Web Server (Score:5, Interesting)

    by Alien54 ( 180860 ) on Sunday April 14, 2002 @12:40PM (#3339175) Journal
    Sadly, this (the fly wed server) is no longer online, but the photos are interesting:

    http://www.conceptlab.com/fly/ [conceptlab.com]

    And yes, there is a video (2.2meg QT) and extended wiring diagrams

  • The cellphone application may have another analogy to biological systems: the setting of priorities of requests.

    You have a weak need to inhale after every exhalation, but that need grows exceedingly pressing -- occupying your entire consciousness, instead of being something of which you're hardly aware -- if you just hold your breath for a minute or so. The priority "inhale" becomes higher than the priority "type the next word in this message", for example.

    For cell-phones, it seems that either there will be too many antennae [wasting resources of subscribers who pay for them, indirectly] or occasionally too few [possibly thwartingg important communication].

    A solution would be a button on the phone that indicates the priority of outgoing calls, or the priority to be assigned for incoming calls. Call me naive, but I think most folks would be happy to turn the priority down a notch, when they are just calling to chat about unimportant fluff.

    I realize that these comments may be more germaine to other applications, since cell-phone communications don't really seem that limited [given the number of folks with those mobiles fixed to their faces], but the idea remains: adding priority to the mix works in nature, and it may work in technology as well.

    • The first time someone is at the grocery and gets a failed outgoing call, they'll just ratchet up the priority and probably never turn it down again.
    • People always believe that what they are doing is always of the highest priority and their cell phones will never go off of the top priority. On the other hand, It might not be a bad idea to do with calls to 911 (here in America) or any emergency service place, but then again, I've almost never gotten the fast busy when I am in range of a tower so I don't see much need for the prioritizing anyways.

      • ...I don't see much need for the prioritizing anyways.

        This issue came up after the Sept 11th WTC and Pentagon attacks. They issued phones to police/fire/rescue that had some sort of priority over normal cell phones. They also claimed that is wouldn't cause normal peoples' phones to drop calls or get fast busy signals (How's that???)

        As for allowing the user to choose the priority of their outbound call, how about having a $1.00/call surcharge for each high priority call made? That'd keep people from leaving the phone of 'high'.

    • I can remember that when AT&T (now Lucent) moved into tho the former Sovjet Union then found a system just like this. Each number was assigned a priority between 1 and 3. 1 was for the party elite, 3 for the "normal" citizens.

      When a prio 1 phone wanted to make a call, a prio 3 call was dropped if there was not enough capacity available. etc etc etc

      This caused the "normal" citizen to keep their lines open for as long as possible, because getting a new connection could be impossible for days ..

      The same will most likely happen here ... and remember .. we all are egoists .. whould you turn your priority down if you know that you next door neighbor does not and get's better reception?
      • this'll sound really stupid, and seem like a lie, you know, the big black bald face kind, but i would, if only because i understand the system.

        of course, you would probably only get voice-mails or text-messaging.

        hmmm, you wouldn't have to "talk" to people.

        to girlfriend: "sorry, i didnt get your call, i was probably at home in bed, thinking of you."

        brownie_points++;
    • I actually think this is a good idea.

      Except-make it server-side QoS instead of client-side, or maybe locked in the phone's firmware not easily accessible.

      You pay for your QoS on top of the minutes. An "emergency" phone could have say 2K minutes but its calls would be dropped on overloaded circuits, so it would cost less. A standard phone wouldn't do anything other than what they do now-too many circuits? Too bad. Already on and the cell fills up? You stay on it. A high-priority plan could cost maybe $50/month more, and would kill off a low-priority phone when needed. This is like the business exec type plan.

      Course people don't like the idea of paying for more importance, but it's sound in business models.
    • Call me naive, but I think most folks would be happy to turn the priority down a notch, when they are just calling to chat about unimportant fluff
      OK, you're naive. Sounds like that voluntary QoS scheme where the Application decides what priority it's packets have on the network. Eventually all applications were marketed as high performance (ie. highest priority) even if they didn't need it. Nobody wanted "inferior" network performance. These ideas are good, until marketing hype gets at it. ATM is good as the App must specify low latency OR high throughput.

      At the end of the day these are artifical measures. What you're actually trying to do is guess the user's priority preference. Linus Torvald's machine would have its CVS repository connection as a high priority and streaming video/Gnutella as a low priority. Warez/Pr0n kiddies will have the opposite QoS preference, however you cannot trust the application to quiz the user about his preference as the application can lie and say it always has highest priority. Hardwiring this decision into the network stack is unfair as it guesses priority (bar kernel recode) but it's the only method that can survive tyranny of the App.

    • Are you insane? Let the customer determine how important their own calls are with relation to the rest of the network?

      Mrs. Smith, the businesswoman with anal retentive tendancies and most of a flagpole lodged up there is going to set her priority at '0' while Mr. Laid Back Construction worker who's calling home to have phone sex with his wife five hours a day will have it at '0'.

      As will the teenybopper calling her friends after school to ask which eyeliner to get at Mervyn's.

      Who, realisticly, doesn't think their calls are more important?

  • How is this news? (Score:2, Insightful)

    by jackb_guppy ( 204733 )
    Is it because some egghead "thought up" a great idea that been around for a while? Or is it that mega company need some stupid press?

    Self adjusting / Self modifing routines have been around "forever". The Fuzzy logic used in some Japnesse clothes washer and air-condioning systems show the way.

    The techinology for the computerized camera positioning of Star Wars 4 (was 1)- The New Hope, had self correcting routines built in.

    So some telecom just thought this up!

    Some news.
    • Star Wars 4 (was 1)- The New Hope

      It was "Episode IV: A New Hope" when it was originally released, even though it was only one movie and was marketed simply as "Star Wars".

  • Natures Contribution? (Score:2, Insightful)

    by todhsals ( 63522 )
    There is nothing new about modeling cellular automata in CS but the natural world could have a lot to contribute in this area. Finding the set of rules that produce a particular equilibrium system can be as much art as it is science. Instead we could look to nature for a pattern that represents our desired outcome. Then look at the rules in the natural system to see how the equilibrium is reached.

    The system would still be fully predictable because the rules can be modeled. If you want to know what will happen if you take a base station off-line just get status info from all the stations, load your model & see how it changes. Need to fix a problem, same solution.

    Perhaps river channel migration can be applied to the problem of network bottlenecks.
    Very cool stuff
  • "We must be willing to give up a certain amount of control, at least of detailed control," in order to let these self-regulating systems succeed.


    err ...brilliant!

  • According to the last sentence of this article, they interviewed T. J. Watson on the subject of autonomous computing. This would interest me greatly, since he's been dead [ibm.com] since 1993.

    Ahhh, conscientious reporting at its finest.
    • To quote the last sentence from the article:

      But Steve White, senior manager of IBM's autonomous computing division at IBM's Thomas J. Watson Research Center, said lower performance shouldn't be an option.

      "I think we should be shooting for autonomous self-administering systems that can do better than humans do," Watson said in an interview.


      Looks to me like they just made a boo-boo, substituting Watson for White.
  • I'm a RF engineer with a major US wireless provider. I do this on a day to day business.

    The configuration of base stations in cell phone networks has always been problematic because you can never predict how many phones will connect to which base station.

    True. However, it doesn't much matter most of the time. We monitor the usage of our sites, and expand those that require it. We also preemtivly expand those that are predicted to require it, and those that we know are going to cover major events, IE concerts, conventions, etc.

    And sometimes adjacent antennas will use the same frequency leading to dropped calls.

    If this is happening, your RF engineer is an idiot. The process of planning what sites use what frequencies is somewhat intensive, but putting the same frequency on two addjacent sites is a complete fsck up. More typical is a site overshooting and interfering with another one several miles away.

    But with dozens of base stations, each broadcasting with six of the 29 available frequencies

    I know that BT has alot more capacity problems than most places here in the US. However, my company uses 3 frequencies per site, from a list of 24. Each site also freqency hops on a list of 18 more freqencies. Hopping really makes all this possible.

    • Say it how it is!

      (/me's a softie at a Northern European telecomms company, and previously spent 6 months analysing base-station logs trying to find significant trends). The comments in the article made me think that the "problems" were made up, as they'd already been solved.

      FP.

      (ignore the sig, the troller account's at 50 too...)

  • that's nice. (Score:2, Funny)

    by ziggles ( 246540 )
    That's great, but what they really need to do is to find a way to keep the fruit flies away from my monitor!

  • But, I thought... (Score:3, Funny)

    by zaren ( 204877 ) <fishrocket@gmail.com> on Sunday April 14, 2002 @02:00PM (#3339395) Journal
    the whole point was to REMOVE bugs from the system...

    -------
    Aww, FSCK! [cafepress.com]
  • So this is like elevators using Fuzzy logic? [ercb.com]. If you everyday work is in a building where the elevators uses Fuzzy Logic, you really notices when you move to a building where they don't. I would have thought that with the rate mobile masts are getting installed everywhere a technology that the article mentions, would already had be created? Now I have no knowledge about how these networks are controlled, but if there is no automated adjustments I can understand why there are so many "dead spots".
  • We all know that fruit flies multiply at an alarming rate. Now consider, in turn, that time flies like an arrow -- if said experiments go awry, it could be discovered that fruit flies like a BANANANANANANANA [tuxedo.org]...
  • Dvorak [pcmag.com] reports on insects and harddrives.
  • As it turns out, autonomous computing is one of the trends we're following closely at O'Reilly. It's one of the major themes at our emerging technology conference (Building the Internet Operating System) in Santa Clara May 13-16. Robert Morris of IBM is giving a keynote on the subject [oreillynet.com], and we've got a whole subtrack on biological models for computation. "Emergence" is also the theme of Steven Johnson's keynote.

    Overall, if you look at some of what's been happening in the peer-to-peer space (with decentralization@yahoogroups.com being a great place to do that), you'll see how all these themes are coming together with the emergence of new internet-scale operating system models.

  • I forgot where I read this, but I remember something saying that the majority of fruit flies are gay :-)! Though, it might be more of nature's form of population control within the fruit fly community.

    Two cats are all that my partner and I really want to deal with, though some of my coworkers are raising kids (two lesbian couples and a gay couple.. I work at a very gay-friendly non-profit). If we were breeders instead, then we'd have too many people on our planet (just a thought).. ;-)

Slashdot Top Deals

Top Ten Things Overheard At The ANSI C Draft Committee Meetings: (5) All right, who's the wiseguy who stuck this trigraph stuff in here?

Close