| The Fifth Generation Mobile Communication System(5G)puts forward the concept of Ultra-Dense Network(UDN),which satisfies the requirements of high regional throughput and concurrent connections by deploying a large number of access points(AP)intensively.But the networking mode of UDN also brings some new challenges,such as timely detection and accurate location of failed AP.Traditional network management methods will incur considerable signal overhead and aren't suitable for UDN.This thesis studies the autonomous healing algorithm in UDN,so as to achieve efficient network failure detection and recovery.The main work of this thesis is as follows:A distributed self-healing mechanism based on cognitive and non-orthogonal multiple access(NOMA)is proposed for wireless access scenario in UDN.A large number of APs are divided into clusters according to the geographic location.Each cluster has a leading-AP(L-AP)to monitor other APs' working status in the cluster,as well as to manage and schedule these APs.The L-AP detects their working status in each cluster,then these information are analyzed based on multi-armed Bandit Model to judge possible AP failure.Once an AP is confirmed to be failed,the impacted users would be connected to the optimal APs in the cluster with the collaboration of other APs.In addition,NOMA is adopted to improve spectral efficiency.Simulation results show that the proposed mechanism could detect the AP failure effectively.Furthermore,the neighboring nodes can ensure the normal rate requirement of the user and improve the system spectrum efficiency when they have sufficient load capacity.A distributed self-healing algorithm based on analytic hierarchy process and load balancing is proposed in the UDN multi-hop backhaul scenario.When a relay fails,the impacted communication nodes(ICDs)collect the status information of normal relays within their detection range,including load,channel quality and residual transmission power of relays,and then generate a candidate relay list for each of the above ICDs.Then,a comprehensive score is calculated for each candidate relay by the analytic hierarchy process.The relay with the highest score is selected as the new next hop relay of the ICD,and the backhaul data is continuously transmitted.In order to improve the robustness of the system,we focus on the balance of the relays' load in the backhaul transmission.Simulation results show that the proposed self-healing algorithm can effectively eliminate the impact of relay failure,and has better load balancing performance than the existing nearest routing algorithms.In the case of multi-relay failure,the user's backhaul data transmission can be better guaranteed. |
