شبکه های بیو حسگر ها: فعال سازی غیرمتمرکز و یادگیری اجتماعی

This paper deals with the dynamics of biosensors and networks of biosensors, where individual biosensors are constructed out of protein molecules. Such biosensors are fully functioning nano-machines. The paper explores methods for decentralized self-activation of networks of biosensors using game-theoretic methods. A global game analysis in terms of a Bayesian game, and a correlated equilibrium analysis is carried out. Also an example of change detection using quickest time detection with social learning is presented. The unifying theme is to understand how local decisions affect global decision making in a multi-agent system.

Biological ion channels are water-filled sub-nano-sized pores formed by protein molecules in the membranes of all living cells [13, 24]. Ion channels in cell membranes play a crucial role in living organisms. They selectively regulate the flow of ions into and out of a cell and regulate the cell’s biochemical activities. In the past few years, there have been enormous strides in understanding of the structurefunction relationships in biological ion channels due to the combined efforts of experimental and computational biophysicists [13, 29]

This paper began with a description of the modeling of a novel ion channel biosensor. We then presented three approaches for analyzing the behavior of networks of biosensors. First, a global game based activation scheme was proposed for analyzing a network of biosensor. We gave conditions under which if each biosensor deploys a simple threshold policy, the global behavior of the network achieves a Bayesian Nash equilibrium. Next, it was shown that if each biosensor unit deploys a stochastic approximation algorithm to minimize its regret, then eventually consensus in decision making is obtained - that is, all sensors pick their strategy from a common convex polytope defined by the correlated equilibrium