Some Optimization Trade-offs in Wireless Network Coding

In this paper, we consider different optimization trade-offs in wireless networks with saturated or possibly emptying packet queues and specify the resulting cross-layer interactions in medium access control (MAC) and network layers. We separately consider scheduled and random access for MAC layer, whereas network layer operations are modeled by network coding or plain routing. Our objective is to analyze the trade-offs among performance objectives of maximizing aggregate or minimum throughput over different source-destination pairs and minimizing (transmission and coding) energy costs using a tandem wireless network model with the assumptions of omnidirectional transmissions (and node costs), interference effects and single transceiver per node. We do not limit the throughput measures to a common minimum value for all source-destination pairs (such as the Max-flow Min-cut value) but specify the entire achievable throughput region that provides the optimization constraints for the general case of multiple sources. We also extend the network optimization problem to the case of possibly emptying queues by specifying the stability region as the constraint set. We consider different multicast communication problems (such as broadcasting and unicasting) and discuss the trade-offs with anycast communication (with arbitrary throughput rates).