Scheduling and performance limits of networks with constantly changing topology

A communication network with tine-varying topology is considered. The network consists of M receivers and N transmitters that, in principle, may access every receiver. An underlying network state process with Markovian statistics is considered that reflects the physical characteristics of the network affecting the link service capacity. The transmissions are scheduled dynamically, based on information about the link capacities and the backlog in the network. The region of achievable throughputs is characterized. A transmission scheduling policy is proposed that utilizes current topology state information and achieves all throughput vectors achievable by any anticipative policy. The changing topology model applies to networks of low-Earth orbit (LEO) satellites, meteor-burst communication networks, and networks with mobile users.