Research On Content Caching And Dissemination Strategy In Wireless Edge Network

In recent years,with the rapid development of mobile Internet and the proliferation of smart devices,continuous growth of mobile traffic has been witnessed.It poses great pressure on mobile network operators from many aspects,such as infrastructure expenditure,operation costs and user experience.To tackle with the challenge,researchers have proposed the proactive networking paradigm.On the one side,machine learning tools have been deeply used in data analysis and user behavior prediction.On the other side,current base stations and mobile devices have been equipped with enhanced computing and storage capabilities.As a result,wireless networks no longer need to follow the reactive pattern of waiting till the advent of traffic requests and responding afterwards.Instead,they can anticipate user demands,cache the contents that may be requested by users in the edge nodes beforehand during off-peak times,and then deliver them to users.Edge caching and content delivery techniques can effectively help decrease redundant network traffic,solve the backhaul bottleneck and save resources with quality of service guaranteed.With the introduction of these two techniques,some new features will arise in terms of resource type and serving mechanism of wireless networks,which makes radio resource management more sophisticated and brings various open issues for researchers to solve.This thesis focus on base station content caching and the dissemination process among users.The main contributions are as follows:Firstly,a distributed cooperative caching algorithm is proposed for backhaul-limited small cell networks.On the one hand,while network densification is the trend for future network,the great number of small cell base stations and their scattered locations have led to the backhaul bottleneck.With content caching involved,the serving rates users can achieve depend not only on wireless channel conditions,the backhaul capacity of base stations,but also the specific contents users request and the caching states of base stations.On the other hand,the overlapping coverage characteristics brings cooperation potential among base stations.In this thesis,the serving rate a user can achieve is derived,with relevant factors considered.A user association strategy is defined to accommodate the scenario,and a simple yet effective caching algorithm is proposed based on potential game.Simulation results and analysis reveal improved delay performance and reduced complexity for the proposed algorithm compared with available schemes.Secondly,considering the incentive design in the dissemination process of a single content,a dynamic incentive mechanism is designed.As a result of incurring costs of mobile users during content sharing,incentives need to be provided.On the one hand,since more users have received the content as the process proceeds,it is not sensible to provide as many incentives as at the beginning.On the other hand,the delivery failure possibility should be taken into consideration during the design of incentive mechanism.In this thesis,user and network behavior in incentive interactions are modeled.It is proved that when both sides are aware of link unstability,the incentive mechanism of providing both reward to a successful delivery and compensation to a failed one is equivalent to a mechanism of only providing a bigger reward to success.The optimal incentives for different network states are derived in two scenarios about information.It can be seen from simulation results the proposed dynamic incentive mechanism can effectively improve network utility.The impact of information asymmetry is analyzed as well.