ASCE: Auto-organización en sistemas de comunicación emergentes – Self-Organization in Emergent Communication Systems TIN2005-09198-C02

This grant studies the behavior of emergent communication systems (communication networks without centralized control, like Internet or ad-hoc networks). We consider systems formed by a number of nodes that communicate and interact locally through links. The nodes route and schedule the transmission of data packets accross the links, to reach a consistent global behavior. Our objective is to ensure that these systems are able to adjust their behavior without a centralized control (hence the term emergent). In this project we make pessimistic assumptions about the system users’ behavior, in order to evaluate the quality of service that the systems can guarantee. To do the analysis under the worst case conditions, we assume the existence of a (bounded) adversary that controls the system. Under this assumption, we evaluate routing and scheduling algorithms, first studying the system stability (impossibility of the adversary to make the number of packets in the system to grow unboundly), and then their performance with respect to other parameters (latency, memory requirements, etc.). For these studies we use models of static networks that we have recently proposed (Continuous Adversarial Queueing Theory, a generalization of the Adversarial Queueing Theory model) and we generalize them for dynamic networks. Moreover, to complement the theoretical studies, we want to compare our analytical results with empirical studies. To do so, we evaluate and compare different algorithms by simulation. Finally, we study the possibility of integrating the algorithms in real networks protocols. Additionally, we want to study strategies that force the system users to respect certain limits in their behavior (for instance, limiting the amount of traffic they generate). To do so, we want to model their interactions with the system as a game and to use Game Theory to analyze it. Our objective is to understand how different rules may influence the users’ strategies, and hence to be able to select the appropriate rules to enforce certain