Research On Coded Caching Technology In Fog Networks

With the explosive development of intelligent devices and the popularity of low latency applications such as video streaming,the current wireless network is facing unprecedented data traffic burst.Coded caching technology stores part of the data in the user terminal during the off peak traffic period,and satisfies the needs of multiple users by the way of coded multicasting during the peak traffic period,which greatly reduces the transmission delay and network congestion,and becomes the main technology trend of the next generation wireless network.With the advantages of distributed architecture and proximity to end users,fog computing has become a promising architecture for future wireless networks.This thesis mainly studies the application of coded caching technology in fog computing network,aiming at further reducing network transmission delay and improving service quality.The main contributions of this thesis are as follows:1.Firstly,this thesis designs a cache system architecture based on the fog computing network and models it as the combination network.In order to solve the problem that the file subpacketization level of optimal coded caching scheme in combination network is too high,this thesis proposes a coded caching scheme based on PDA.In this scheme,according to the PDA array,the server places MDS-encoded packages of files in the cache memories of relays and users,and generates the coded multicasting information to meet the needs of users.Then this thesis applies the PDA with priority to the combination network with resolvability property,and implements the caching scheme supporting user priority.Finally,this thesis proves that the performance of the coded caching scheme based on PDA can significantly reduce the file subpacketization level while the transmission rate is close to the optimal scheme.2.The combination network with high symmetry is limited in practical use.Therefore,the architecture of the fog network cache system is modeled as a more general two-hop network.Firstly,this thesis proposes a centralized coded caching scheme for two-hop network.Based on the MDS-encoded files,the scheme guarantees the coded multicasting opportunity in the delivery phase by carefully designing the cache content of relays and users.Then,this thesis analyzes the transmission rate and normalized transmission time of the scheme,and summarizes the “optimal” network topology and the “worst” network topology of the scheme in the case of parallel transmission and successive transmission respectively.The simulation results show that the centralized coded caching scheme proposed in this thesis for two-hop network is the best caching scheme at present.3.In order to increase the flexibility and applicability of the caching scheme,this thesis proposes a decentralized coded caching scheme for two-hop network.In this scheme,each user caches part of the MDS-encoded packages of files randomly and independently,and the server creates multicasting opportunity according to the cache content of relays and users in the delivery phase.The independence of user cache content makes it unnecessary for the server to know the identities and number of actual active users in the placement phase,which ensures the flexibility of the scheme.Then this thesis designs an online decentralized coded caching scheme for two-hop network.This scheme ensures that the cache content of relays and users is consistent with the files in the server by updating the server files and updating the cache content of relays and users,so as to maintain the validity of the cache content.Finally,this thesis summarizes the “optimal” network topology and the “worst” network topology of the scheme in the case of parallel transmission and successive transmission respectively.The simulation results show that the decentralized scheme proposed for two-hop network in this thesis has the best performance in the caching schemes based on random cache placement.In addition,the performance of the decentralized scheme proposed for two-hop network is very close to the optimal centralized scheme,while the decentralized scheme has higher flexibility and applicability.