Decentralized optimal control in shared resource pools of video service nodes

Abstract

Tri-network convergence as one of the basic feature, the new generation of national information infrastructure is the major strategic decision to realize the leaping development of national information industry, modern service industry and traditional industries. Business needs of large-scale video concurrent access and user-oriented providing of high-quality services will be the new challenges of application access system in the future tri-network convergence. In the domain of video service network in tri-network convergence, we focus on the problem of optimal resource allocation of RSPE (Real-time Service Provide Equipment) nodes in shared resource pools. In this paper, we propose a fully distributed optimal control solution, named RSPE-DOC. RSPE-DOC treats underlying physical servers as a shared resource pool, and RSPE nodes as resource consumers. RSPE-DOC implements the information exchanging strategy and distributed optimization method among RSPE nodes, leading to a distributed optimal control framework. Experimental results indicate that RSPE-DOC is able to ensure the convergence of dissemination process and find the optimal resource allocation, if the RSPE nodes are not changed substantially in resource management process.