On The Key Techniques And Optimization Of Coded Caching

The explosive growth of mobile networks in terms of business scale and data traffic causes severe challenges for wireless communication technologies.The most important core of wire-less communication techniques is how to fully explore the potential network resources to achieve high spectral/energy/cost-efficient communication systems.Furthermore,in order to alleviate the heavy burden of network traffic,how to break through the bottleneck of the conventional techniques,and fully exploit and integrate the available network resources,becomes one of the most significant goals of designing the next generation networks.Recently,the cache-aided communication technology has shown its advantages in reducing network energy consumption and data traffic,and improving the network spectrum efficiency.Therefore,the cache-aided communication becomes one of the key techniques of the next gen-eration networks,and is widely introduced to achieve efficient communication systems.On the basis,the coded caching that integrates the benefits of cache-aided communication and net-work coding techniques,could further improve the utilization of storage resource to alleviate the requirements for traditional communication resources.This thesis aims to exploit the coded caching to realize the tradeoff between the storage and communication resources,such that the transmission issue of the explosive network traffic can be addressed.To achieve lower latency and higher energy efficiency,this thesis focuses on the joint optimization of content caching and delivery within different network constraints,to fully realize the local cache gain and coded multicasting gain in the coded caching schemes.Firstly,this thesis overviews the research progress and key techniques of the applicable coded caching schemes.So far,the index overhead including placement overhead and deliv-ery overhead,which is caused by file partition,is studied.Accordingly,a unified framework to analyze the index overhead is proposed,and then is investigated in the decentralized coded caching schemes with finite subpacketization.The trade-off among subpacketization,achiev-able performance,and index overhead is revealed,which shades some lights on the optimal subpacketization design.Secondly,to reveal the impact of heterogeneous file sizes on the realization of coded mul-ticasting gain,this thesis proposes a file partitioning and grouping based coded caching scheme,which could effectively cope with the difference of file segments caused by the file size het-erogeneity.By optimizing the file caching proportions,the proposed scheme can maintain the coded multicasting opportunities among users,so as to approach the optimal performance of the centralized scheme in some certain scenarios.Thirdly,the coded caching scheme for a wireless multicast network with asymmetric links is studied.A parallel delivery strategy is accordingly proposed based on the superposition cod-ing and successive interference cancellation techniques,in which the wireless links are logically divided into multiple parallel partially shared links.It is unveiled that the coded multicasting gain and transmission efficiency can be simultaneously realized by optimizing the transmission load,power and rate.In addition,the minimum transmission time and the fairness requirement among users can be guaranteed as well.Lastly,the energy consumption model of coded caching scheme in wireless networks is explored,in which the energy consumptions in both file placement and delivery phases are in-vestigated.Then a file partition-based coded caching scheme is proposed for D2D networks,in which the file placement,D2D transmission and BS transmission strategies are j ointly optimized to minimize the overall average energy consumption.The analysis reveals the relationship be-tween the overall energy consumption and the optimal coded caching scheme design.Above all,this thesis studies the coded caching technique,and reveals the impact of dif-ferent and practical network characteristics on the realized coded multicasting gain.The place-ment and delivery strategies are jointly optimized to achieve the tradeoff between the storage and communication resources.