Characterize how causality-based inferred topology approximates the true wireless network

Determine in what ways the network topology inferred by applying causality-based methods such as Granger causality and transfer entropy to node transmission time series approximates the actual wireless communication graph, specifying the structural aspects of the true network that are captured by the inferred network.

Background

The paper surveys prior work on wireless network topology inference that detects links by computing causality measures between node-level time series, notably Granger causality and transfer entropy. Although these approaches often achieve reasonable accuracy in simulations, their theoretical underpinnings and the exact nature of the structures they recover are not well established.

The authors explicitly note uncertainty about how the topology produced by such causality-based inference relates to the true underlying network. This motivates their global, Markov-chain-based estimation framework, which provides operator-norm guarantees, but the relationship for standard causality methods remains an open question.