In the 5G Heterogeneous Ultra Dense Network environment, due to the dense deployment of the network and the diversity of user service types, a more flexible network selection algorithm is required to reduce the network blocking rate and improve the user’s quality of service (QoS). Considering the QoS requirements and preferences of the users, a network selection algorithm based on Dueling-DDQN is proposed by using deep reinforcement learning, which defines the state, action space and reward function of the different services to maximize the cumulative reward value of the network selection algorithm. The simulation results show that compared with other algorithms, the proposed algorithm can effectively reduce the network blocking rate while reducing the switching times.
In the paper, a hybrid clustering routing strategy is proposed for railway emergency ad hoc network, when GSM-R base stations are destroyed or some terminals (or nodes) are far from the signal coverage. In this case, the cluster-head (CH) election procedure is invoked on-demand, which takes into consideration the degree difference from the ideal degree, relative clustering stability, the sum of distance between the node and it's one-hop neighbors, consumed power, node type and node mobility. For the clustering forming, the weights for the CH election parameters are allocated rationally by rough set theory. The hybrid weighted-based clustering routing (HWBCR) strategy is designed for railway emergency communication scene, which aims to get a good trade-off between the computation costs and performances. The simulation platform is constructed to evaluate the performance of our strategy in terms of the average end-to-end delay, packet loss ratio, routing overhead and average throughput. The results, by comparing with the railway communication QoS index, reveal that our strategy is suitable for transmitting dispatching voice and data between train and ground, when the train speed is less than 220km/h.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.