Approximation Algorithms for Bandwidth and Storage Allocation Problems under Real Time Constraints

Abstract

The problem we consider is motivated by allocating bandwidth slots to communication requests on a satellite channel under real time constraints. Accepted requests must be scheduled on nonintersecting rectangles in the time/bandwidth Cartesian space with the goal of maximizing the benefit obtained from accepted requests. This problem turns out to be equal to the maximization version of the well known Dynamic Storage Allocation problem when storage size is limited and requests must be accommodated within a prescribed time interval. We present constant approximation algorithms for the problem introduced in this paper using as a basic step the solution of a fractional Linear Programming formulation. This problem has been independently studied by Bar-Noy et al [BNBYF+00] with different techniques. Our approach gives an improved approximation ratio for the problem.