Research On Spectrum Access Technologies In Wireless Communication Networks

With the continuous growth of mobile terminal equipment,the demand for spectrum resources of wireless communication services continues to expand,resulting in an occurrence of spectrum access congestion worldwide.At the same time,according to the United States Federal Communications Commission,the utilization rate of the frequency band below 3GHz is between 15% and 85%,which is uneven.Efficient and flexible spectrum access methods are one of the key technologies to solve the problems of "spectrum access congestion" and "low spectrum utilization".To this end,this article will study the spectrum access technology in wireless communication networks,including the random access process in mobile communication networks and dynamic spectrum access in cognitive radio networks.This article first aims at solving the phenomenon of "spectrum access congestion",and thus studies the optimization of random access procedure in mobile communication networks.The random access procedure is the first step for the user equipment to initiate spectrum access to the cell base station.The user equipment implements uplink synchronization and obtains physical uplink channel through the random access procedure.In the case of dense traffic,random access channels(RACH)congestion is prone to occur due to preamble collisions.In order to solve this kind of RACH congestion,based on the existing random access process,this article proposes a novel random access mechanism based on real-time access intensity detection(RAID-RA).This mechanism includes two phases of detection and resolution.Competitive users can sense the real-time random access intensity of the current time slot during the detection phase,and calculate the back-off probability according to the preamble interference power during the resolution phase.The simulation results show that compared with the existing random access mechanism,the RAID-RA mechanism has a random access throughput gain of 10%-50% under different random access traffic and different user distribution conditions.A certain performance improvement is also achieved in the success rate,the number of attempts,and the delay.In addition,in order to meet the differentiated performance requirements of different random access traffic,this article proposes a priority-based PARAID-RA mechanism based on RAID-RA.This mechanism implements differentiated random access performance through power ramping strategies and back-off probability differentiation.The simulation results show that under the condition of dense random access traffic,the PARAID-RA mechanism can guarantee the success rate and delay of high-priority random access types without affecting the performance of other random access types.This article then studies the problem of dynamic spectrum access decision-making in cognitive radio networks in response to the phenomenon of "low spectrum utilization".In cognitive radio networks,secondary users increase the utilization of authorized spectrum by dynamically sensing and utilizing vacant sub-channels.Multiple channels and multiple users dynamic spectrum access can be modeled as a partially observable stochastic game problem.In order to solve the problem of observable random countermeasures in this part,based on deep reinforcement learning knowledge,this article proposes a deep Q networks based noncooperative dynamic spectrum access(DQN-NCDSA)algorithm and a deep Q networks based cooperative dynamic spectrum access(DQN-CDSA)algorithm.The simulation results under the non-zero-sum game dynamic spectrum sensing scenario show that compared with the random strategy and the greedy strategy,the two proposed algorithms show different degrees of gain in the detection rate and discovery rate of vacant channels.In considering the exchange of sensing information of adjacent secondary users,the DQN-CDSA algorithm is better than the DQN-NCDSA algorithm in the vacant channel detection rate.On the other hand,in the zero-sum game dynamic spectrum exploitation scenario,we consider two reward design criteria:individual reward maximization and network reward maximization.Simulation results show that the reward criteria based on individual reward maximization performs better.At the same time,simulation results also show that compared with the random strategy and greedy strategy,the two proposed algorithms have a 25% performance improvement in channel capacity and vacant channel utilization.