The goal of this research is to learn from biological models to structure, design and implement robust and scalable ambient intelligent systems.
Biological models have a number of interesting properties that can be useful in the design of robust and large-scale ambient intelligent systems. An example class of biological models is based on stigmergy [4,5]. Stigmergy is used by ant and termite colonies to communicate indirectly via pheromones deposited in their environment. Based on this principle ants and termintes coordinate their actions with one another. This manner of communication has a number of advantages:
The mechanism of coordination is fully decoupled from the entities (ants and termintes) such that if an entity fails the colony can still proceed with its task. Hence, the communication mechanism is robust with respect to failures.
The communication between entities is local. In other words, entities can only communicate when they are in the same environment. Several researchers [1,2,3] have argued that locality is key in the design of scalable ambient intelligent systems.
Other biological models exhibit similar properties with respect to scalability and robustness and should be considered as well.
AmbientTalk, in its current incarnation, has a weak notion of proximity embedded within, based on the wireless communication range. Proximity is important in many biological models in order to support the concept of locality. In a first step the goal is to extend AmbientTalk with a richer notion of proximity such that the semantics of local communication can be adapted to the application at hand.
Next, we will experiment with a number of biologically-inspired models such as the stigmergy briefly explained above and investigate what programming constructs are necessary to easily express programs based on these models.
 Swarm Intelligence, ISBN: 0-19-513159-2
 Emergence: The Connected Lives of Ants, Brains, Cities, and Software, ISBN: 0-684-86875-X