On Throughput Optimality With Delayed Network-State Information

The problem of routing/scheduling in a wireless network with partial/delayed network (channel and queue) state information (NSI) is studied in this paper. Two cases are considered: (i) centralized routing/scheduling, where a central controller obtains heterogeneously delayed information from each of the nodes (thus, the controller has NSI with different delays from different nodes), and makes routing/scheduling decisions; (ii) decentralized routing/scheduling, where each node makes a decision based on its current channel and queue states along with homogeneous delayed NSI from other nodes. For each of the cases (with additional flow restrictions for the decentralized routing/scheduling case), the optimal network throughput regions are characterized under the above described NSI models and it is shown that the throughput regions shrinks with the increase of delay. Further, channel and queue length based routing/scheduling algorithms that achieve the above throughput regions are proposed in this paper.