As the number of concurrent users of network applications grows, it becomes increasingly difficult to provide services reliably through conventional centralized methods on the Internet. In other words, nearly everyone will one day use the Internet all the time for a variety of reasons and current methods wont be able to handle this. The Bio-Networking architecture (Bionet) is being developed as a solution to this problem by Tatsuya Suda and his team at the University of California, Irvine. It employs biological principles, especially ones derived from emergent systems such as insect colonies. Bionet applications are implemented using mobile agents called Cyber-Entities that scurry about network nodes and provide network services.
Applications using Bionet should be:
-
Scalable- They must be able to handle enormous user demand (billions of network nodes and users).
-
Adaptable- They must be able to handle constantly changing user demand and network conditions.
-
Survivable- They must be able to handle network failures.
Emergent Biological systems such as ant colonies show all the above properties. They consist of very simple (aka dumb) individual units such as ants who might not even know what they are doing. Together, however, they are able to pull off some remarkably complex emergent behavior. They can get together to successfully forage for food, ward off predators and recover from damage to their colonies. In all their behavior, these colonies employ elegant algorithms that are emulated in Bionet. For instance, the ant colony algorithm can be used to find shortest paths between the nodes of a network.
Bionet-based applications and CEs are analogous to ant colonies and individual ants. Ants wander out of their colonies, obtain food and bring it back. CEs forage around the network, looking for users who want their services in return for energy. This energy is used by CEs to reproduce and migrate. CEs ideally migrate and reproduce in a manner that ensures optimal populations and distributions over networks according to the prevailing user demand and distribution.CEs can co-operate with each other through mechanisms such as pheromone emission to produce complex behaviors in a decentralized way.
Bionet is both a paradigm and middleware for creating network applications. Bionet-based applications can be used to produce any number of network applications.Researchers have already created Aphid, a web server, using it.The project shows a lot of promise not only to achieve its own goals, but also to demonstrate the power of biological algorithms.
The Bio-net Simulator can be downloaded for some hands-on experience with the architecture.