— The idea of the cognitive network, drawn from previous research in cognitive radios and cross-layer design, is an attempt to address issues of complexity and scalability by allowing the network to self-adapt and reach end-to-end goals. This paper identifies three critical design decisions that must be considered when creating a cognitive network: the behavior of the decision-making elements, amount of computational state available to these elements, and the degree of control they have. To measure the effect of these design tradeoffs, we develop and define a metric called the “price of a feature.” This metric is used to quantify the effect of these properties on a cognitive network with the objective of maximizing the lifetime of a multicast flow in a wireless environment. For this cognitive network, we show that the price of selfish behavior and partial control are more significant than the price of ignorance.
Ryan W. Thomas, Luiz A. DaSilva, Madhav V. Marathe