Research On Resource Allocation In Wireless D2D Communication System

With the rapid increase in the number of mobile smart devices and people’s demand for wireless communication,the problem of insufficient system capacity and shortage of spectrum resources is gradually emerging.Device-to-Device(D2D)communication is one of the key technologies of the 5th Generation Mobile Communication Technology(5G),which is based on cellular facilities to improve resource utilization,data transfer rates and user throughput.However,when D2 D users multiplex the spectrum resources of cellular users,co-channel interference problems can arise.Therefore,for D2 D networks,a rational resource allocation strategy is of great importance.In addition,Relay-auxiliary D2 D Communications have been proposed as a complement to traditional D2 D communications to increase system capacity for LongTerm Evolution.Therefore,this paper addresses issues related to D2 D communication systems and Relay-auxiliary D2 D Communication,the main elements of which are summarised below:Firstly,for the D2 D communication system,the resource optimization problem of single-cell uplink communication networks is studied.The aim is to maximise the sum rate for users while maintaining the quality of service for communications over the cellular link and the D2 D link.Because the optimisation objective of D2 D communication is a non-convex function,it is difficult to obtain a global optimum,and for this reason the optimisation problem is considered to be split into two steps,namely mode selection and resource allocation.First,a K-nearest neighbour query algorithm is used to select the corresponding communication mode for the D2 D user based on the user’s geographical location;then the Asynchronous Advantage Actor-Critic(A3C)based algorithm is used to allocate resource blocks and power to different D2 D users for the multiplexed communication mode.Unlike traditional algorithms,this algorithm treats each D2 D user as an agent and uses a centralised training distributed execution approach to rationally allocate resource blocks and power.Simulation results show that the proposed scheme outperforms other methods in terms of learning efficiency and network performance,and can effectively reduce inter-system interference between D2 D users and cellular users,maximising the sum rate for users.Secondly,due to the limited communication distance between D2 D users,the introduction of Relay-auxiliary D2 D Communication is considered.To further improve the performance of the whole network,a resource allocation scheme for joint relaying with multi-agent hybrid deep reinforcement learning is proposed with the objective of maximising system throughput,taking into account the uncertainty of channel state information.By using the Deep Q network method to allocate the relay and resource blocks,and using the Deep Deterministic Policy Gradient algorithm to allocate the transmit power.Simulation results show that the introduction of Relayauxiliary D2 D Communications significantly outperforms conventional D2 D.The convergence,stability and accessibility of the proposed method are demonstrated by comparison with other methods.