In this paper, we present a multi-gate mesh network architecture that has been developed to ensure high performance and reliability under emergency conditions when a system expects to receive power outage notifications and exchanges. In order to handle the metering traffic, we introduce a back-pressure based scheduling algorithm, which takes into account both the hop-count, as well as the queue length of each mesh node. We first present the objective function of the scheduling algorithm in a centralized manner. We then present a practical implementation of the packet-scheduling scheme. The simulation results indicate that under the context of the multi-gateway network, our novel packet-scheduling scheme can indeed signi-ficantly improve the network's reliability, self-healing, through-put and delay performance.