Self-sizing networks: local vs. global control

We consider the problem of extending the network "self-sizing" framework to locally controlled networks, in which resource allocation decisions are made at the node level. Schemes for self-sizing and adaptive resource optimization have been proposed in the past for globally controlled networks. We show that by performing online resource allocation at each node based on their local knowledge, we can achieve considerable bandwidth savings, congestion reduction and also satisfy QoS at the packet level. Online traffic measurement based on effective bandwidths has been used as the tool for estimating real time bandwidth allocation, which implicitly guarantees the QoS. Simulation results also suggest that by making some of the nodes aware of their neighbors resource availability, higher self-sizing gains can be attained.