Research On Heuristic Resource Allocation Algorithms In Mixed Service Scenario

The continuous network evolution and the increasing demand for business lead to tremendous pressure and challenges in resource allocation and usage.Although the resource capacity of hardware devices is getting larger,it could not satisfy the ever-increasing and various user requirements,and constituting reasonable resource allocation algorithms is an effective way to solve these problems.Most of the traditional resource allocation algorithms take the single resource utilization as an optimization objective,ignoring the diversity of service requests and resource types in the network.Service providers sometimes need to handle multiple different types of service requests simultaneously,and differentially allocate network resources to these services.However,most existing resource allocation studies only consider one type of service request,or distinguish multiple types of service requests to provide services separately,without considering the impact among different types of service requests.On the basis of introducing the resource reservation model and summarizing the research status of resource allocation algorithms,this thesis focuses on the resource allocation problem in mixed scenario of two and three different types of services,including Immediate Reservation(IR)service,Pre-Scheduled Reservation(PSR)service,and Malleable Reservation(MR)service.The thesis first gives its research background and significance,introduces the definition of three reserved services of IR,PSR and MR to be studied,and summarizes the characteristics and reservation mechanism of each service.It also analyzes the resource allocation problem in mixed scenario of two and three different services,and briefly describes the research status at home and abroad.The second chapter summarizes the research basis of resource allocation in mixed service scenario,mainly including the introduction the principle and application of heuristic algorithm,summarizing the application scenarios of resource reservation and network resource classification,and analysis on the solving algorithms for multi-resource joint assignment problem.For the bandwidth resource allocation problem with PSR and MR mixed services,the third chapter gives the relevant parameter constraints,and establishes an optimization model aiming at minimizing the rejection ratio of service requests.Two heuristic algorithms based on efficient transmission strategy are proposed to solve the model,which are entitled as the Resource Allocation based on Fixed Transmission Strategy(RAFTS)algorithm and the Resource Allocation based on Segmentation Transmission Strategy(RASTS)algorithm.The proposed RASTS algorithm selectively changes the path according to the resource status of the K shortest candidate paths in each time interval,so as to make full use of the dispersed resources in the network.In addition,two transmission rate thresholds are set to control the number of path changings,which can reduce the overhead caused by frequent path changings.The performance of the proposed two algorithms is compared by simulation experiments.The results show that the RASTS algorithm is better than the RAFTS algorithm in reducing the service requests rejection ratio and improving the network resource utilization.To make the study conform to the complex actual network environment better,the fourth chapter focuses on the resource allocation problem in the scenario with three mixed types of services.It is assumed that there are three types of services: IR,PSR,and MR,and the resource types requested by each service request are also expanded into compute,storage,and bandwidth.In this mixed services environment,an optimization model with the optimization objective of maximum service profit is established according to the unit revenue and cost of resource.Then a Joint Resource Allocation based on Compute,Storage and Bandwidth(JRACSB)algorithm is proposed to solve the model.We use Matlab tools for algorithm programming.The simulation experimental results of algorithm performance show that compared with RAFTS algorithm,the proposed JRACSB algorithm could effectively improve service profit and reduce the rejection ratio of service requests.The fifth chapter summarizes the work done in this thesis,and gives the further work that will be carried out in the aspects of network scenarios,algorithm efficiency,resource revenue and cost.It also presents possible future research directions.