| As the number of users increases, the traffic in the Internet has increased exponentially. Cooperative caching has been considered as a solution to reduce Internet traffic. One component of cooperative caching is the placement of objects, which determines where the objects are stored. This thesis investigates the placement of streaming multimedia objects in cooperative caching, with the objective of maximizing the bandwidth observed by users.; Streaming multimedia objects are usually resizable by using different compression schemes. Using this property, the placement problem is formulated as a nonconvex quadratically constrained quadratic program (QCQP). An existing global optimization technique, the Global Optimum Search (GOS), is applied to the QCQP, leading to two linear programs, which are solved iteratively to determine the final placement. Simulation results on a system based on a 6-node topology and systems based on the Sprint IP backbone network show that the bandwidth observed by users increased drastically. |
