Data Download And Repair Strategy For D2D Distributed Storage Nteworks

With the diversification of mobile data service and rapid growth of traffic in recent years,the load pressure on existing mobile communication system during peak traffic periods is continually increasing.The communication model forwarded by the base station has been unable to meet the communication demand of large-capacity,low-latency,low-energy of the massive mobile data service in the future.The intelligent computing and storage capacity enhancement of smart devices makes distributed storage system based on Device-to-Device(D2D)communicationa key opportunity to solve the problem of reliable storage and download of massive mobile data.This thesis focuses on the mobility of the devices in D2 D distributed storage systems,to reduce data download and repair cost transmission costs(ie communication overhead)as research goals,with the redundant error-tolerant strategy such as replication,maximum distance separable(MDS)code,and regenerating code.It is a propose to build a D2 D link strong availability and low costwireless data storage and repair network system.The influences ofenvironment and caching scheme parameters on system communication cost is analyzed in detail,and the optimal repair interval under different parameters and scenarios is obtained Furthermore,a scheme integrated replication and MDS code,named double replication MDS(DR-MDS)is proposed,which improve the repair cost and download cost of MDS code in D2 D distributed storage network.The main content of this thesis is organized around the following aspects:Firstly,the thesis studies the D2 D distributed storage system model and analyzes the communication cost.For the random process of nodes entering and leaving the system,the data is periodically repaired to maintain the availability of the D2 D link..Under the condition of instantaneous repair,compare the effects of different schemes such as replication,MDS code and regenerating code and related parameters on communication cost,and draw that the communication cost of 2-replication scheme is the smallest,under the condition of periodic repair,the communication cost and the maximum repair interval of different cache schemes are compared.It is found that the maximum repair interval is related tofault tolerance of the coding scheme.The higher the fault tolerance,the larger the range of the maximum repair interval and choose appropriate caching scheme can reduce communication cost.Secondly,the thesis focuses on the impact of impact of D2 D link availability and repair interval on D2 D distributed storage system.The impact analysis of the download locallity proves that the number of connected nodes is smaller,the greater the probability of transmitting data through the D2 D communication link,and the lower the communication cost.When request frequency is high,MDS code is more advantageous in terms of D2 D link availability and communication cost,and there is an optimal repairinterval,which can be derived from the D2 D repair success probability and the D2 D download success probability.Finally,to solve the large bandwidth problem for MDS codes,a double replication MDS storage strategy combining MDS and replication scheme is proposed.The scheme improves the system performance by backing upencoded data,which not only reduces the repair bandwidth of a single node,but also improves the availability of D2 D link,which greatly reduces the repair and download cost,and expands the application of redundant fault-tolerant repair strategy in D2 D distributed storage network.