Distributed Energy Aware Cross-Layer Resource Allocation in Wireless Networks

In this paper, we consider the joint scheduling, routing and congestion control mechanism in [4] while incorporating a comprehensive physical layer model that considers both primary half-duplex constraints and the power-SINR-rate relation, and heterogeneous nodal power budgets. We consider a cross-layer scheme comprising of a primal-dual congestion controller and an energy aware back-pressure (EABP) scheduler that decides routing, scheduling, power and link rate selection based on the queue length information as well as an excess energy consumption state at each node. The handling of nodal power constraints in our scheme is essentially the same as that in [9] and [6]. For completeness, we provide a self-contained proof that the cross-layer scheme asymptotically achieves optimal fair allocation of the network resources. Then this scheme is used to motivate the design of a scalable and implementable distributed slow time-scale (DSTS) power control algorithm, which can be combined with rate adaptation and known distributed link scheduling algorithms to approximate the centralized EABP scheduler. In this way, we provide a candidate solution to complete the network utility maximization (NUM) based protocol stack for multi-hop wireless networks. We provide simulation results that show what are potential performance gains.