Complex Systems, Self-Organisation, and Simulation

Many of the systems we build are complicated, with lots of moving parts and interactions. We have developed engineering techniques to address these complications (for example layered and nested abstractions) that break-up large systems into smaller component parts that can be developed and reasoned about independently. But an increasing number of engineered systems are complex, and resistant to this kind of decomposition. Complex systems don't have easily-identified components, and exhibit “emergent” behaviours that self-organise in unexpected ways out of small-scale interactions. Paradoxically this can make systems more predictable, more responsive, more robust, and more stable, using self-organisation to respond to even small changes in the environment in ways that maintain desirable macroscopic behaviour. How should we study complex systems, and introduce self-organisation more widely into engineering? In this tutorial we'll explore how to think about complexity, drawing on insights from physics, biology, mathematics, and computer science. We'll illustrate the ideas using communications networks, which exhibit a lot of complex systems features really well, and use simulation to answer a question about how communications systems fail using complex systems ideas. And we'll look forward to new techniques emerging in the literature that we can potentially apply to engineering self-organising systems.