The emerging interest in cognitive networks, smart grids, and self-organizing networks is motivating heightened research on collaborative processing strategies that enable networks to learn and respond to information in real-time. Adaptive networks are well-suited to perform decentralized information processing and decentralized inference tasks. They are also well-suited to model self-organizing behavior such as animal flocking and swarming. These networks avoid centralized processing and perform in-network inference and control decisions without relying on omnipotent agents (or fusion centers). This is because solutions that rely on information fusion are not scalable, are hard to adapt to changing network conditions, and create single points of vulnerability and information bottlenecks.
Adaptive networks consist of spatially distributed agents that are linked together through a connection topology. The topology may vary with time and the nodes may also move. The agents cooperate with each other through local interactions and by means of in-network processing. The diffusion of information across the network results in various forms of self-organizing behavior and collective intelligence. A key property of adaptive networks is that all agents behave in an isotropic manner and are assumed to have similar abilities. This kind of behavior is common in many socio-economic and life and biological networks where no single agent is in command.
This talk describes recent developments in distributed processing over adaptive networks and illustrates the techniques by studying self-organization in biological networks such as bird formations, fish schooling, bee swarming, and bacteria motility.