A local greedy scheduling scheme with provable performance guarantee

In recent years, there have been many efforts to develop low-complexity scheduling schemes that can approximate optimal performance in multi-hop wireless networks. A centralized sub-optimal scheduling policy, called Greedy Maximal Scheduling (GMS) is a good candidate because it achieves high throughput. However, its distributed realization requires O(|V|) complexity, which becomes a major obstacle for practical implementation, where |V| is the number of nodes in the network. In this paper, we develop a simple distributed scheduling policy for multi-hop wireless networks. It achieves O(log |V|) complexity by relaxing the global ordering requirement of GMS. Instead, it deterministically schedules only links that have the largest queue length among their local neighbors. We show that, it still guarantees a fraction of the optimal performance, which is no smaller than GMS. We also further improve its performance and address some important implementation issues. The simulation results confirm that the new scheduling scheme achieves the performance equivalent of GMS and significantly outperforms state-of-the-art distributed random access scheduling policies,