The objective of the proposed research is focused on low probability of detection (LPD) applications by minimizing energy transmission and transmission in directions not advantageous to intended receivers based on cognitive networking techniques via multi-hop routing, adaptive power control, and directional antenna command and control loops. The goal is to develop distributed cognitive algorithms operating in cognitive nodes to intelligently maintain throughput and connectivity while accounting for time varying mission requirements. Additional time and location dependencies are considered such as the position of observers. Geographic position data from the network is used to predict changes in the topology and estimate the multi-objective costs of muli-hop paths in the network. An emphasis is placed on fast approximations to ensure real-time implementation. The set of Pareto-optimal paths are found based on the estimated costs, allowing the node to select the best path for any cost function.