Research On Transmission Load And Overhead Optimization In Coded Caching Schemes

Mobile users in wireless network have strong patterns on time in daily activities,resulting in high time-varying of network traffic.Coded caching,first proposed by Maddah-Ali and Niesen,is a network coding technology,which can transfer the pressure of network traffic from peak times to off-peak times and relieve the pressure of network traffic during peak times.Based on the centralized network model,the first coded caching scheme was proposed by Maddah-Ali and Niesen(MN scheme),achieving the order-optimal transmission load with uncoded placement when users request different files.PDA(Placement Delivery Array)is a class of combinatorial construction,which is used to describe the centralized coded caching scheme called combinatorial coded caching scheme.Linear coded caching scheme consists of caching matrix,coding matrix and decoding matrix,which can be used to characterize the centralized coded caching scheme from the view point of linear algebra.A special multiaccess coded caching scheme can be derived by a method of transformation from PDA.In this paper,we study the general optimization of transmission load in the centralized coded caching schemes when the users request the same file,and the overhead optimization in the multiaccess coded caching scheme transformed from the MN PDA.Research results of this thesis are as follows.1.Based on the centralized network model,we study the graph optimization in the combinatorial coded caching schemes.Given a PDA and any user request d,a PDA-d graph can be obtained according to certain rules: the vertices and edges of the graph represent the signals broadcast by the server and the relationships among them,respectively.By designing a search algorithm for the graph,we can find the most redundant signals,which can be retrieved by other signals.Then,the server eliminates the redundant signals,reducing the transmission load in the wireless channel.2.Based on the centralized network model,we study the rank optimization in the linear coded caching schemes.In the delivery phase,when users request the same file,we recharacterizes the linear coded caching scheme and transforms its coding matrix,making it associated with users request.By calculating the rank of all users' coding matrix,the maximum linearly independent number of the signals broadcasted by the server can be derived,then the lower bound of transmission load is obtained.3.Based on the multiaccess network model,we study the multiaccess coded caching scheme with overhead.In the multiaccess coded caching scheme transformed from the MN PDA,assume that there exists access overhead when users access cache-nodes and broadcast overhead when server broadcasts signals.We calculate the total overhead of the scheme,and establish the mathematical model of overhead optimization by a superposition method.In this paper,graph optimization in the combinatorial coded caching scheme and rank optimization in the linear coded caching scheme,are both general optimizations in the centralized coded caching schemes,reducing the transmission load;the multiaccess coded caching scheme with overhead transformed from the MN PDA is characterized,and the mathematical model of overhead optimization is established by a superposition method.