Link Scheduling under Correlated Shadowing

We study the effects of stochastic shadowing on scheduling in wireless networks. Previous work has generally assumed that shadowing affects different signals independently. Concentrating on “compact” networks, where very little or no spatial reuse is possible, such as in indoor environments, we form a model of correlation between the shadowing components for different signals. We analyze two fundamental measures: the maximum number of simultaneous transmitting links, and the fewest time/frequency slots or channels needed to schedule all the links. Based on the correlation model, we characterize (up to constant factors) how these measures scale with correlation strength and the number of links. We also give nearly optimal algorithms to compute such schedules, as well as to optimize the maximum weighted sum of simultaneously transmitting links. The latter can as well be extended to arbitrary sets of similar length links, under a suitable model of correlations, by partitioning them into compact subsets.