Abstract: Abstract: In order to schedule a gateway node in wireless sensor networks, this paper proposes a new gateway node scheduling method which based on a Markov chain. The research method of this paper is: first, theory analyses one single network node's existential state, and its access behavior; second, we base on the gateway and terminal node's common status, via single node's state transition probability relation to build up system's balance equation; third, according to this balance equation, anode's scheduling is executed, as well as a description of the system working process; finally, the system's performance parameters of the scheduling method is estimated. This paper's experiment built a wireless sensor network simulated model, set up the simulation environment, constructed several gateway nodes, terminal nodes, wireless channel bandwidth, node's transmitting power, channel's signal-to-noise ratio, Rayleigh flat fading channel, etc. system parameters. Simulation analysis was conducted on this network model's bandwidth utilization and forced interrupt probability; then changing a node's signal-to-noise ratio's value, the simulation compared network capacity and blocking probability performance. Research result: first, with the increasing of a user's call rate, the bandwidth utilization also grows, because the call's increasing inevitable causes a system's collaboration gateway node to increase. When the system reaches a steady state, the bandwidth utilization is maintained at 10% to 70%. Second, the forced interrupt probability is also increased with a user's increasing call rate, on the one hand, call's increase leads a gateway node's state transition frequency to increase, on the other hand, the system tendency to a steady-state probability is also increased, thus interrupt probability increases, but the experimental results derived interrupt probability value is always less than 0.1, and this can content general quality of service (QoS) performance requirements. Third, system capacity is increased with call volumes increasing. With call increasing, the system spectrum utilization is correspondingly increased, and this will inevitably lead to the growth of system capacity. Meanwhile, when the call volume grows to a certain extent, the system capacity will gradually flatten, which is due to the system interrupt probability that will also flatten with call's increasing. The system will eventually reach a state of equilibrium finally. In order to enhance a node's efficiency in a relay system, it is necessary to do a reasonable arrangement for the access node itself working condition and between each node. According to a node's own state transition situation and interaction between each node in relay system, a relay system node scheduling model based on a discrete time Markov chain model is constructed. System equilibrium state is analyzed. The results show that, based on a relay node's rational utilization, we improve system bandwidth utilization, while keeping a lower system interrupt probability. It can improve the stability and robustness of farmland nursery irrigation sensor networks.