Cooperative and Non-Cooperative Resource Sharing in Networks: A Delay Perspective.

Abstract-From multi-description/multi-path routing for multi-media applications to content distribution in P2P networks, to community networking, many forms of resource sharing have been proposed to improve network performance. From the perspective of any single user, when ignoring the interaction among users, all such schemes reduce to various forms of providing parallelism and, hence, increased throughput. Focusing on parallelism is by no means sufficient as it ignores the existence of many users with potentially similar strategies. In this paper, we focus instead on the delay performance of a multi-user system where resources are shared. We illustrate the benefit, in an average delay sense, of resource sharing among many (potentially strategic) users via a multiqueue multi-server problem. We use a fork.join queuing model to provide analytical results in a special case of homogeneous users and servers. Our proposed model is simplistic, and yet, it does capture the trade-off between parallelism, traffic load increase, and reassembly/synchronization delay to a large extent. Furthermore, we prove the robustness of a certain locally optimal strategy to non-cooperation in a Nash equilibrium sense. [ABSTRACT FROM AUTHOR]