Nature Inspired Systems

Transcription

1 Nature Inspired Systems Mark Shackleton Intelligent Systems Lab BTexact, Adastral Park, UK 12th April 2002

2 Overview of this presentation BTexact Intelligent Systems Lab Nature Inspired Systems why? Example1: FlyPhones - dynamic frequency allocation in mobile phone networks, inspired by biological development Example 2: ADDICT - adaptive defense of computer networks, inspired by an immune-system analogy Conclusions BTexact Technologies

3 Intelligent Systems Lab BTexact s AI research capability, based at Adastral Park, UK Key technology areas: 85+ patents in these areas - soft / fuzzy computing; software agents - intelligent data analysis; profiling & personalisation - optimisation, scheduling & resource planning - intelligent business systems - nature-inspired approaches adaptive, evolutionary & decentralised systems BTexact Technologies

4 Nature Inspired Systems - why? Our current repertoire of system design and management tools appears to be inadequate for dealing with the complexity of IT today Fact: nature has evolved a variety of strategies to cope with dynamic, highly unpredictable constraints. Working hypothesis: these strategies can inspire suitable (if not perfect) solutions to a variety of complex problems. BTexact Technologies

5 Nature Inspired Systems - why? Natural systems exhibit desirable properties: - adapt to wide-ranging unpredictable changes to their environment - resilient: self-healing, self-protecting, autonomous - highly decentralised, complex heterogeneous architectures that work! But there are also consequences: - Controllability - we must give up some control - Precision - good enough rather than optimal - Comprehensibility - why did that happen? Complex Systems analysis can help BTexact Technologies

6 FlyPhones

7 Bio-inspired solutions at work: the FlyPhones project Cell differentiation in the fruit-fly (bristles distribution). Frequency allocation in mobile phone networks. Q: What could those problems have in common? A: The way nature solved the first can help us deal with the second. BTexact Technologies

8 Frequency allocation - the problem Interference between neighboring base stations should be prevented. The number of available frequencies is limited. Bandwidth has to be very carefully distributed between adjacent cells. BTexact Technologies

9 Cell differentiation for fruit flies Most cells have the potential to make bristles. They all start to express the corresponding gene (greyscale = density of the associated transcription factor) But they gradually specialise until only a few actually develop a bristle. BTexact Technologies

10 Underlying mechanism This process obeys a fully decentralised control mechanism. It involves a local positive feedback loop Coupled with crossinhibition of neighboring cells (Delta-Notch signalling) BTexact Technologies

11 Why is an original solution needed? This is in fact a very complex problem: - The base stations are not regularly distributed (i.e. they don t have the same size and/or number of neighbours) - The continuously fluctuating traffic must be taken into account. A centralised decision process is not particularly well adapted But a system allocating frequencies on the basis of local competition between cells is. BTexact Technologies

12 The Flyphones algorithm The equivalent of the natural feedback loop is implemented for each available frequency. Through the negotiation process, each base station starts to develop a preference for some frequencies (moving away from the unstable equilibrium) And simultaneously inhibits its neighbors from using them. BTexact Technologies

13 Applying FlyPhones Combat Net Radio - battlefield deployment lines of communication, 250 channels, mobile transceivers - currently solved centrally, statically (obvious difficulties!) - FlyPhones can solve it too - contract research to use FlyPhones in dynamic management Comparing Different Types of Frequency Constraints Co-Site + Far-Site Constraint Co-Site + Far-Site + Spurious Co-Site + Far-Site + Spurious + Intermod Time/s BTexact Technologies

14 ADDICT - Adaptive - Dynamic - Defense - for Information & Communication Technologies

15 The disappearing perimeter In 1996 and there was a notion of the trusted and untrusted network.... at that time, there was a clear perimeter between... the anarchic, hacker-ridden Internet and the carefully controlled internal network. No more. Leon Erlanger 21st Century Security, Internet World Dec BTexact Technologies

16 The disappearing perimeter (2) BTexact Technologies

17 The paranoid approach Should we consider keeping every single piece of hardware on full-time security alert? Do we really have to choose between seamless connectivity and security? No, but... - Ad-hoc networks and firewalls don t like each other! - However, this is more because of the way we think about security than because of some fundamental incompatibility. BTexact Technologies

18 Goals and inspiration There are several biological models of adaptive defense: - Self-amplified intrusion alert in social insects. - Inflammatory response of the immune system. A good security system: - automatically closes dangerous backdoors, - leaves other (safe) passageways conveniently open, - isn t a permanent hassle for the legitimate user! BTexact Technologies

19 The inflammatory metaphor explained The paranoid approach is like maintaining your immune system on such high alert that you re permanently running a fever, just to be ready in case infection strikes! Because this isn t energy efficient, life has evolved ways of detecting where and when a defensive barrier must be raised. This is (more or less) what we call the inflammatory response. BTexact Technologies

20 ADDICT algorithm dx dt ( 1 x) n x = N n + α N i=1 x i βx BTexact Technologies

21 Epidemic history No defense No defence No Static defence defense Static defence A.D.D.I.C.T. Fraction infected Time BTexact Technologies

22 The ADDICTed office Intruder (red in blue) Private area (green & blue) Infected node High alert (bright halo) Low alert (faint halo) Private area (red) Public area BTexact Technologies

23 Autonomic..? Self-managing and Self-configuring Self-healing (FlyPhones) Self-protecting (ADDICT) Self-optimizing (or, dynamically adaptive) Scalable (by virtue of local interactions) There is less need for central control, or for human intervention However, a human can define constraints BTexact Technologies

24 Conclusions Nature-inspired approaches can bring valuable design advantages This doesn t mean copying nature, rather using a similar logic Nature-inspired approaches should only be used when they map reasonably well to the problem Another highly relevant discipline is that of Complex Systems Expect nature-inspired systems soon, but don t necessarily expect to know it! BTexact Technologies

25 Acknowledgements Richard Tateson - creator of the FlyPhones algorithms and software Fabrice Saffre - creator or the ADDICT algorithms and simulations For further details please - mark.shackleton@bt.com BTexact Technologies