Research On Intelligent Routing Of SDN-based Air-ground Integrated Network

The Intelligent Transportation System(ITS)becomes popular and provides various convenient services for vehicle drivers.It is expected to improve safety and comfort during driving.All these services and applications are based on network access.However,the great number and dynamic of vehicles bring a significant challenge for networking.Due to the fast and flexible mobility,Unmanned Aerial Vehicles(UAVs),i.e.drones,are widely used in communication of the complex environment in recent years.It is deployed as a wireless relay in the air,which plays a vital role in the future air-ground integrated network.Nevertheless,the existing researches only focus on the improvements of routing protocols,it is failed to fully utilize the advanced mobility of UAV.Firstly,to address various deficiencies of urban vehicular network,this paper proposes an SDN-based and drones-assisted vehicular network routing model,which is an air-ground integrated model for networking.For a given position,it evaluates the demand for drone relay according to the distribution of terrestrial vehicles.Subsequently,the drones with network models are dispatched to most rational positions to provide the best relay nodes for routing.Considering the time consumption of computation and deployment,we also design an anti-inertial prediction of future distribution of vehicles to deploy in advance.More,because multiple drones are scheduled in real-time,it should locate diverse solutions with restrictions of time and distance.However,conventional optimization technologies are powerless and inefficient for these requirements.Inspired by the migratory behavior of the nomadic tribe,this paper proposes a novel intelligent optimization algorithm,namely Nomad Algorithm(NA),and takes self-learning and multimodal enhancement to handle the above problem to be optimized.NA mimics the nomad tribe which always migrates to the more livable place.Members of the tribe are divided into herdsman or scouts with different search strategies,which corresponding to local and global search,respectively.Based on the coexistence of multiple tribes on the grassland,this paper also designs the multiple optimization strategies of NA.Compared with other conventional algorithms,this succinct algorithm is efficiency and adaptive with great advances in convergence speed and accuracy.Applied on routing optimization of the aforementioned integrated network,our NA can quickly find the fittest locations for drones that assist the terrestrial vehicular network routing in the best measure.Finally,the simulation of the SDN-based air-ground integrated network which is optimized by intelligent NA has been conducted based on the real-world map and floating car data of Chengdu City,China.Simulation results denote that The PDR and throughput achieve 5%-15% improvements,meanwhile,the average number of hops is reduced by about 2.It can be considered as a flexible,efficient,and promising scheme for networking.