Edge Server Placement Strategy Under Multi-conditional Constraints

Edge computing is one of the important pillars of future information technology.It plays an important role in many scenarios,such as smart cities,smart homes and intelligent transportation systems.In recent years,the research on edge computing has become a hot topic gradually.People hope to use a new architecture to make computing tasks run on mobile devices,and these computing tasks usually require a lot of calculations.These portable devices include mobile phones,tablet computers and some wearable devices.However,these devices have weak processor capabilities,and they are limited by their own power sources.These constraints determine that these large-scale computing tasks cannot be run on these devices directly.In recent years,the research on edge computing has been continuously improved,and it is considered to be a good solution to these problems.The main idea of edge computing is to put these large-scale computing tasks on the edge servers which close to the users.However,there are few researches on the locations of edge servers.Everyone thinks that the edge server has been deployed in an ideal location to meet the various needs of edge computing.However,the selection of edge servers locations has always been a relatively difficult problem.Therefore,this thesis proposes two strategies for edge servers locations to further promote the development of edge computing.The main work of this thesis is as follows:(1)An edge server location strategy based on cost-effectiveness is proposed.When we study the edge server deployment solution,from the perspective of cost savings,we choose to deploy the edge server on an existing base station instead of re-finding a new location.But when we observe the base stations in a certain city,we find that the base stations in some places are deployed too densely.Therefore,we do not need to deploy an edge server on each base station,because if this happens,the edge server will have a large overlapping areas,which will waste a lot of resources.Our method is divided into two stages.In the first stage,we adopt the method of dynamic programming,giving each base station a cost of deploying an edge server,and then describing each process of deploying as an recursive formula.In the second stage,we regard the coverage area of each edge server as a circular area,and then summarize all the intersecting situations of circular areas.We use the mathematical geometry method to find the total area of these areas,and then substitute all of them in the recursive formula.The comparisons with greedy,full permutation and randomized algorithms prove the effectiveness of our method.(2)An edge server location strategy based on multi-objective optimization is proposed.In this thesis,We propose the method of the deployments of edge servers with the computing load,network latency,and costs.First,we consider the computing load and network delay encountered during working,and then combine these factors with the deployment of the edge server.The number of users and the distances between them are also related.These two factors are converted into an objective function.Then,the relationships between edge servers,base stations and users are considered.Thus some constraints are obtained.Then we use a weighting method to transform a multi-objective problem to a single-objective one,which can be solved easily,and then a preliminary deployment plan is obtained.Finally,we consider the deployment costs of the edge servers.We divide each edge server into different clusters,then give each edge server a cost constraint function,and each cluster has a total constraint function.We use greedy and simulated annealing algorithms to minimize the numbers of edge servers.