Adaptive bandwidth allocation control for Virtual Paths in broadband networks

This thesis focuses on studying the fundamental tradeoff between processing and transmission costs. Our research objective is to develop algorithms for Virtual Path (VP) bandwidth allocation that achieve an optimal balance between the two costs, while keeping the implementation complexity low. This work presents four implementation specific algorithms for VP bandwidth allocation that operate in a distributed fashion. Based on a threshold scheme the size of VP is dynamically adjusted to reflect changes in the VP occupancy; the rate of change is restricted by processing resources. The framework and algorithms were evaluated by simulations. These algorithms resulted in efficient performance. When the relative cost between the transmission and processing resources is low, the least complex algorithm performs as well as those of the more complex algorithms.