Researches On Key Techniques For D2D Communications In Heterogeneous Networks

With the explosive increasement of mobile user equipments,the scarcity of spectrum resource becomes more serious and the burden of base station(BS)is also getting heavier.Meanwhile,the proximity-based communication resides in each place.As a result,D2D communication receives great attentions and becomes one of the promising technologies in 5G networks due to its advantages in proximity-based service area.Comparing with the traditional cellular communication,D2D communication has the capability of offloading traffics from BS and core network,as well as decreasing the power consumption and transmission delay and increasing the system throughtput.D2D communication is regarded as an important technology for proximity-based service which allows direct communication between two devices via sharing the same spectrum resource with cellular users.In this paper,we aim at solving several key technical issues including architecture,device discovery and access procedure,mode selection,resource allocation and transmission capacity via effective design and optimization in heterogenous networks consisting of cellualr users and D2D users.The main contributions of this thesis can be concluded as follows:Firstly,one hybrid communication architecture including both Outband and Inband mode is designed based on WiFi-Direct and Bluetooth working on ISM as well as D2D working on licensed spectrum.In the proposed architecture,an clustering algorithm is proposed based on Bluetooth interface.Comparing with the existing clustering algorithm,the proposed algorithm increases the cluster life time while guaranteeing the throughput.In addition,we propose the transmission scheme that users inside the cluster transmit or receive via WiFi-Direct interface.Otherwise,transmission between each cluster head can be carried out on licensed spectrum.Furthermore,we also study and design two different schemes of device discovery and access procedure which are initialized by UEs and S-GW,respectively.System model based on Markov process is presented and the performance analysis shows the viability in terms of cluster life time,system throughput and power consumption.Secondly,the technology of mode selection and resource allocation is investigated.Generally,mode selection or resource allocation is considered individually in most of the existing literatures.In addtition,although the joint mode selection and resource allocation is also investigated in some studies,only few D2D pairs and cellular users are involved.The main difference from the existing researches is that we investigate joint mode selection and resource allocation in large scale D2D networks by utilizing coalitional game theory.Firstly,a coalitional game algorithm based on the constraint of guard zone is proposed.Unlike the traditional coalitional game,two members are involved simultaneously in each round such that it increases the convergence rate of the algorithm.Moreover,we carry out the derivation and analysis of the stability,convergence rate and radius of guard zone for the proposed algorithm.Finally,we propose one distribution method based on the stochastic geometry model.Simulation results show the reliability and efficiency of the proposed algorithm.Thirdly,we study the transmission capacity for multi-mode D2D communication by utilizing stochastic geometry.In this paper,underlay D2D communication is categorized into two different sharing modes including single D2D pair sharing and multiple D2D pairs sharing.We first analyze transmission capacity for both of them and hence obtain the relationship between the density of cellular users and D2D users corresponding to both of the two different sharing scenarios.In addition,we derive the transmission capacity for D2D communication in overlay mode.Furthermore,we provide the optimal strategy of relay selection for relay-assisted D2D communication under the maximum optimization constraint of transmission capacity.Considering the fact that multiple optimal relay nodes may exist,we finally make the further optimization based on further constraint of minimum energy consumption.