Research On Resource Management Algorithms In SDN-based Novel Heterogeneous Ad-hoc Networks

With the development of mobile communications technology,cloud computing technology and artificial intelligence(AI)technology,an increasing number of enhanced applications which is computing-intensive and delay-sensitive have emerged.Limited by the computing resources and storage capabilities of mobile terminals and network nodes,as well as the high service latency of cloud computing,traditional wireless heterogeneous networks and Ad-Hoc networks cannot provide effective support for these applications.Novel heterogeneous Ad-Hoc networks based on software defined network(SDN),which integrates new technologies such as SDN and fog computing,involving hybrid heterogeneous network integration,resource management of networks,dynamic networking and intelligent routing,etc.,provides new research ideas for satisfying the enhanced requirements of applications.Focusing on resource management issues,SDN-based multi-agent cooperative resource sharing(MCRS)algorithm and SDN-based mobile-aware joint offloading and resource allocation(MJORA)algorithm are proposed in this thesis and the proposed algorithms are analyzed and verified by computer simulation.The main work of this thesis is as follows.(1)SDN-based multi-agent cooperative resource sharing algorithm is proposed.This algorithm is to achieve resource sharing between fog nodes by multi-agent cooperation,aiming at minimizing the average service latency of the network.The algorithm,combining the theoretical framework of the coalitional graph game,initializes the cooperation matrix and utility function at the first,and then continuously updates the cooperation strategy in the iterative process to the final convergence.In each iteration,the optimal proportion of the cooperation between the current fog node and other fog nodes is calculated,and the increment in the utility function of the corresponding coalition is calculated,and the node with the largest increment in the utility function is selected for forming a coalition.In this algorithm,the SDN control layer is responsible for collecting and updating global state information and scheduling the distributed cooperation between fog nodes.Simulation results demonstrate that MCRS algorithm can balance the workloads of different fog nodes,reduce the average service latency significantly compared with the case without cooperation.Besides,the result obtained by the MCRS algorithm is close to the global optimal solution and has lower computational complexity.(2)SDN-based mobile-aware joint offloading and resource allocation algorithm is proposed.The algorithm comprehensively considers the mobility of users,task offloading,computing resource allocation,and bandwidth resource allocation,minimizing the average service latency of users in the network while satisfying mobility constraints,energy constraints,and limited resource constraints.In this algorithm,the SDN control layer provides centralized control.The algorithm is divided into offloading matching process and resource optimization process.In the offloading matching process,the offloading matching sub-problem is solved by a matching game algorithm,and the offloading matrix is obtained,that is a set of pairs between user equipment and fog nodes.Based on these pairs,the resource optimization sub-problem is further decomposed into independent sub-problems.Then the Karush-Kuhn-Tucker(KKT)condition and the alternate optimization method are used to solve the resource optimization sub-problem,and finally the solution of the joint offloading and resource allocation problem is obtained.Simulation results show that compared with other benchmark algorithms,the average service latency of the MJORA algorithm is significantly lower and the support for mobility is better.