Research On Routing Mechanism Of Cognitive UAV Swarm In Flying Ad Hoc Networks For Wide Area Emergency Communications

For wide-area emergency communications,most of the existing methods are based on terrestrial infrastructure,such as terrestrial base stations,for resource scheduling,which is limited by space and environment to a certain extent and lacks flexibility.UAVs open up new avenues for natural disaster management and applications due to their inherent maneuverability and flexibility.In order to solve the contradiction between the rapid growth of user equipment and the shortage of spectrum resources,and a single UAV cannot meet the multi-mission requirements,the search and wireless coverage functions are provided for the ground by combining cognitive radio and UAV swarm coordination.Most base stations fail to function properly in disaster scenarios,therefore,without the support of centralized infrastructure,cognitive UAV swarm transmits the collected information through multi-hop routing in the form of a centerless distributed ad hoc network.In the routing process,cognitive UAV swarm has some problems such as node position changes dynamically,energy limitation and dynamic spectrum selection,and path stability problems such as link interruption during transmission.In this context,single-path and multi-path routing strategies in cognitive UAV swarm are studied.The main research results of this thesis are as follows:(1)This thesis proposes a maximum utility intelligent routing strategy based on Qlearning.A novel three-hierarchical aerial-ground integrated network architecture is established for wide-area emergency communication scenarios,and cognitive radio is applied to UAV swarm.In order to obtain the maximum utility routing,a Q-learning framework for cognitive UAV swarm is established,a routing algorithm based on Qlearning is proposed,and the state set including UAV nodes,positions and the remaining energy is defined.A utility function including the reachable rate between nodes and the remaining energy of nodes is designed.In order to reduce the computational complexity of the algorithm,a sector area constrained by position,arc and direction is defined,and a candidate UAV selection strategy constrained by the central angle and signal-to-noise ratio in the sector area is further designed.A reward function is designed by jointly considering routing metric and candidate UAV selection strategy.To accelerate the convergence rate,a dynamic learning rate is introduced.The simulation results show that the accumulated rewards and convergence speed of the system adopting the dynamic learning rate are higher than those of the fixed learning rate and decaying learning rate.(2)This thesis proposes a multi-path stability routing strategy based on hypergraph matching.An air-ground integrated network architecture is established,and the aerial cognitive UAV swarm is divided into the structure of multiple spherical regions.The stability of UAV nodes is designed by jointly considering the inter-regional node dynamics,link quality and distance,and a stability optimization function including node stability and channel availability probability is established.By transforming the 0-1integer linear programming problem of node selection and channel selection into the hypergraph matching problem,a multi-dimensional hypergraph matching multi-path routing algorithm based is proposed.A vertex set including UAVs and channels is defined,a hyperedge set connecting limited nodes and channels in each region is designed,and the weight of hyperedges including stability and channel availability probability is further designed.The simulation results show that,compared with the greedy algorithm,the multi-dimensional hypergraph matching algorithm that continuously finds the optimal solution through local search can obtain multiple node disjoint paths and achieve higher path stability.