Online Resource Allocation Problems∗

In this paper, we consider an online resource allocation problem, where buyers with limited budgets are interested in purchasing items that consume a limited amount of resources. In general, there is no nontrivial performance guarantee for the worst cases. However, after imposing two reasonable restrictions, we can propose two online algorithms, a greedy algorithm and a primal dual algorithm, with non-trivial performance guarantees. We further show that these two algorithms are asymptotically the best among all algorithms, either deterministic or randomized, in terms of competitive ratios.