Computational aspects of optimal schedules with power control and multiple bitrates selection in multihop wireless networks

A significant amount of research on optimal scheduling has focused on the case where links can transmit with only a single bit-rate and a single transmission power. This paper develops techniques to accommodate multiple bit-rates and multiple transmission powers as well as a continuum of transmission powers. Optimizing over a large number of bit-rates and transmission powers typically increases the computed throughput at the expense of increasing the computation time. This paper examines the trade-off between computed throughput and computation time, with emphasis on the case when the relationship between bit-rate and SINR is the same as it is for 802.11a/g. One finding of this paper is a progression of Pareto optimal schemes that increases the computed capacity. A second finding is that optimizing over two transmission powers and two or three bit-rates achieves nearly the same throughput as optimizing over a continuum of transmission powers and all supported bit-rates. However, optimizing over a continuum of transmission powers and all supported bit-rates is far more computationally complex than optimizing over two transmission powers and a few bit-rates.