Tractable Scheduling Algorithms for Self-Backhaul in 5G Networks

We consider the problem of scheduling 5G wireless networks with self-backhaul. The key difficulty is coordinating the transmissions on the backhaul and access links since a node cannot transmit and receive at the same time. A well-known solution to this problem is "utility maximization with backpressure". However, this suffers from a complexity issue in that a single scheduling entity needs to know the channel conditions on all links whose traffic passes through a single node. We therefore propose two simpler algorithms in which the scheduler at a node only needs to know the channel conditions on its adjacent links. These scheduling algorithms utilize a master/slave concept in which the backhaul links make their decisions first and then the access links make their decisions based on the decisions made by the backhaul. We demonstrate that these simplified schedulers only create a small performance loss for a natural example with random node placements. We also examine the extent to which the interference conditions at a node fluctuate due to the changes in transmission patterns. This in turn indicates the potential from carrying out scheduling on a network-wide basis.