Research On Performance And Energy Threshold Optimization For LTE-LAA Under Imperfect Spectrum Sensing

With the increase of smart terminals and their business applications,the global mobile traffic increases rapidly,and further the licensed spectrum for mobile communication systems becomes tenser and tenser.Some researchers thus propose the cellular communications technology to operate on the unlicensed spectrum,which is called licensed assisted access(LAA).In order to achieve the harmonious coexistence between the cellular system and the Wi-Fi system on the unlicensed band,the LAA scheme stipulates that small base stations(SBSs)need to confirm the state of the channel before data transmission,and that SBSs can only send data when the channel is detected to be idle.In the LAA scheme,SBSs use energy detection technology to detect the channel.However,due to the inherent limitations of the energy detection technology,the channel detection results may be affected by the channel environment,the energy threshold and etc,resulting in imperfect spectrum sensing.At present,in the field of the impact of imperfect spectrum sensing on coexistence performance,little research has been done.Therefore,this thesis mainly studies the impact of imperfect spectrum sensing on coexistence performance and the optimization of energy threshold.Firstly,considering the possible imperfect spectrum sensing in the channel detection process,this thesis models the 3GPP LAA scheme as a new two-dimensional discrete time Markov chain,where the time scale in the backoff stage is redefined.Based on the proposed model,the expression of collision probability and throughput is derived,and further the impact of imperfect spectrum sensing on coexistence performance is deeply analyzed.In order to analyze the coexistence performance by comparisons,this thesis also studies the impact of imperfect spectrum sensing on the performance of different backoff mechanism based coexistence network.The simulation results show that the imperfect spectrum sensing of the channel can greatly affect the throughput and fairness of the coexistence network.Secondly,considering the fact that the energy threshold can affect the accuracy of channel detection and further the coexistence performance,this thesis proposes a Q-Learning based energy threshold optimization algorithm.In this algorithm,SBSs are acted as the agent,the energy threshold is defined as the action,the different combinations of fairness and throughput are defined as the state,and the agents obtain corresponding rewards by performing different actions.Based on the proposed Q-Learning framework,the agent can continuously interact with the environment untill it learns the optimal energy threshold,which makes the the coexistence network achieve the maximum throughput on the premise of ensuring the fairness.Simulation results show that the Q-Learning based dynamic energy threshold optimization scheme is superior to the traditional fixed energy threshold scheme.