Hierarchical scheduling algorithms with throughput guarantees and low delay

Abstract

We propose distributed scheduling algorithms that guarantee a constant fraction of the maximum throughput for typical wireless topologies, and have O(1) delay and complexity in the network size. Our algorithms resolve collisions among pairs of conflicting nodes by assigning a master-slave hierarchy. When the master-slave hierarchy is chosen randomly, our algorithm matches the throughput performance of the maximal scheduling policies, with a complexity and delay that do not scale with network size. When the master-slave hierarchy is chosen based on the network topology, the throughput performance of our algorithm is characterized by a parameter of the conflict graph called the master-interference degree. For commonly used conflict graph topologies, our results lead to the best known throughput guarantees among the algorithms that have O(1) delay and complexity. Numerical results indicate that our algorithms out-perform the existing O(1) complexity algorithms like Q-CSMA.