Experimental design of unmanned aerial vehicle channel selection for emergency scenarios in underground spaces
[Objective]Recently,underground space disasters have occurred with increasing frequency,highlighting the urgent need for a comprehensive,multidimensional support system for emergency communication.A key task in emergency communication for underground spaces is the rapid reorganization of communication networks to relay disaster information in real time.Unmanned aerial vehicles(UAVs),with their flexible deployment,unmanned operation,and all-terrain mobility,can play a vital role in emergency rescue efforts.In the event of disasters such as fires,collapses,or floods,UAVs can quickly access affected areas and establish stable ad-hoc networks using their onboard communication devices.UAVs can promptly transmit the collected disaster information back to the rescue center,thereby enhancing the efficiency of rescue operations.However,while extensive research has been conducted on UAV applications in low-altitude areas,further development is necessary for their use in underground spaces.Particularly,there is a lack of research on experimental platforms that integrate scientific research and teaching.This gap has resulted in a disconnect between the educational content and real-world scenarios.Therefore,establishing a scenario-based teaching and practice platform for emergency communication networks in underground spaces is essential.By enhancing practical teaching content,we can effectively guide students in learning modern emergency communication network technologies,ensuring a strong connection between theoretical instruction and real-world applications.[Methods]In this paper,we employ hypergraph theory to abstract the rapidly changing network topology into a dynamically evolving hypergraph,modeling the evolution of interference relations through the insertion and deletion of superedges.Specifically,we reformulate the dynamic channel selection problem in UAV communication networks as a hypergraph coloring problem.To adapt to the dynamic network topology,we leverage reinforcement learning theory to enhance the traditional hypergraph coloring algorithm and develop a dynamic hypergraph coloring algorithm.Additionally,when the topology of the UAV communication network changes,the traditional greedy algorithm requires recalculation at each time slot,whereas the Q-learning algorithm can quickly converge based on prior learning.Building on this discussion,we propose a dynamic channel selection algorithm based on the Q-learning framework,which can adjust the learning rate parameters according to the rate of topological changes.[Results]To validate the effectiveness of our proposed algorithm,we conduct extensive simulation experiments to observe how the proportion of link conflicts in the network changes with varying algorithm iterations and UAV flight speeds.Additionally,we analyze the changes in average throughput across different network sizes,comparing the Q-learning algorithm to the greedy algorithm.The simulation results demonstrate that our algorithm can dynamically adjust the learning rate parameters based on the topological change rate of the UAV communication networks,achieving an optimal trade-off between algorithm performance and convergence speed.And the network performance of this algorithm is better than that of the greedy algorithm.[Conclusions]In conclusion,we utilize hypergraph theory to design an adaptive channel selection algorithm based on the Q-learning framework and establish a simulation experiment platform using ns-3 simulation software.Specifically,we employ the ns-3 platform to create a network simulation experiment for an underground space emergency scenario,helping students apply their theoretical knowledge of communication networks to practical engineering contexts.This work lays a solid foundation for developing a simulation teaching platform for UAV emergency communication networks within the interdisciplinary integration of electronic information science and emergency management.
emergency communication in underground spaceUAV communicationns-3 simulationresource allocationQ-learning
万方数据
维普
