Routing and Power Allocation for MIMO-Based Ad Hoc Networks with Dirty Paper Coding

Recently, researchers showed that "dirty paper coding" (DPC) achieves the capacity region of MIMO Gaussian broadcast channels (MIMO-BC). So far, there has been little study on how this fundamental information-theoretic result will impact the cross-layer design for MIMO-based ad hoc networks. To fill this gap, we consider the problem of jointly optimizing DPC power allocation at the physical layer and multihop/multipath routing at the network layer for MIMO-based ad hoc networks. This optimization problem turns out to be a challenging non-convex problem. To address this difficulty, we transform the original problem to an equivalent problem by exploiting the uplink-downlink duality. For the transformed problem, we propose a solution procedure that integrates Lagrangian dual decomposition, conjugate gradient projection based on matrix differential calculus, and cutting-plane methods.