The accuracy of short-term traffic flow prediction is one of the important issues in the construction of smart cities, and it is an effective way to solve the problem of traffic congestion. Most previous studies could not effectively mine the potential relationship between the temporal and spatial dimensions of traffic data flow. Due to the large variability in the traffic flow data of road conditions, we analyzed it with “dynamic”, using a dynamic-aware graph neural network model for the hidden relationships between space-time in the deep learning segment. In this paper, we propose a dynamic perceptual graph neural network model for the temporal and spatial hidden relationships of deep learning segments. This model mixes temporal features and spatial features with graphs and expresses them. The temporal features and spatial features are connected to each other to learn potential relationships, so as to more accurately predict the traffic speed in the future time period, we performed experiments on real data sets and compared with some baseline models. The experiments show that the method proposed in this paper has certain advantages.

The secret key rate is one of the main obstacles to the practical application of continuous-variable quantum key distribution (CVQKD). In this paper, we propose a multiplexing scheme to increase the secret key rate of the CVQKD system with orbital angular momentum (OAM). The propagation characteristics of a typical vortex beam, involving the Laguerre–Gaussian (LG) beam, are analyzed in an atmospheric channel for the Kolmogorov turbulence model. Discrete modulation is utilized to extend the maximal transmission distance. We show the effect of the transmittance of the beam over the turbulent channel on the secret key rate and the transmission distance. Numerical simulations indicate that the OAM multiplexing scheme can improve the performance of the CVQKD system and hence has potential use for practical high-rate quantum communications.

With the explosive growth of multimedia services and the continuous emergence of new space tasks, the spatial task scheduling timeliness problem is of great concern. The high computational cost of existing task scheduling methods is not suitable for the time-varying scenarios of space-based networks. This paper proposes a scheduling optimization method containing an atomic task offloading model based on maximum flow theory and a dynamic caching model. Firstly, the model calculates the task offloading upper limit in the satellite network based on the maximum flow theory to achieve the maximum volume of offloaded tasks to improve the resource utilization of idle satellites. Then, we design onboard task offloading and buffer optimization algorithms to reduce the request load of single-satellite atomic tasks. The method improves the overall computational performance and timeliness of the satellite network and reduces the waiting time of atomic tasks competing for resources. Finally, we analyze the time complexity of the proposed method and construct a simulation experiment scenario. The performance comparison results with various baseline models show that the proposed method has certain time complexity and task execution timeliness advantages.