Research On Several Key Technologies Of Congestion Management In Heterogeneous Wireless Networks

With the development of the information society,the wireless access systems designed for specific types of traffic have been unable to meet the users increasingly complex and various traffic demands.The future wireless communication network is the integration of heterogeneous wireless networks,through which the wireless resources can be more effectively used.The diversity of heterogeneous wireless networks and the rapid growth of the network data will lead to network congestion more easily,which affects the overall network performance.Therefore,the congestion management mechanism is the key technical problem in the complex heterogeneous networks.Based on the research of all kinds of principles and technologies of congestion managements in wireless networks,a deep research of congestion management is made with regard to access side,backhaul network and fronthaul network in this paper.The key points are access control algorithm in access side,active queue management algorithm in backhaul network and data compression algorithm in fronthaul network.With regard to the congestion produced in access side,an access control algorithm based on Q learning is proposed in heterogeneous wireless networks consisting of LTE,Femto and D2D.JRMM controller can select the appropriate network to access and allocate resources for each arriving call according to effective network coverage,network load status,different traffic types,terminal mobility,D2D mode and so on by using the return value reflecting unity after accessing network.Simulation results show that the proposed algorithm can achieve better resource allocation and higher system gain while ensure low call blocking probability.With regard to the congestion produced in backhaul network,QRED algorithm based on RED is proposed in order to achieve practicality and low complexity.The proposed algorithm based on Q learning can relieve the defect of parameter sensitivity of RED by adaptively adjusting the maximum packet dropping probability.According to the dynamic network environment,the optimal strategy is obtained by learning,which will improve the network performance and avoid producing network congestion.With regard to the congestion produced in fronthaul network,a data compression algorithm based on discrete sine transform proposed in distributed base station architecture.The time domain signal is transformed by discrete sine transform according to the characteristics of the LTE baseband signal,and the coefficient after conversion is divided into two blocks in accordance with the energy concentration characteristics.Numbers of bit are allocated in different blocks and Lloyd-Max quantizer is used to quantify the coefficient in each block.Finally,the compression ratio is improved by Huffman coding under the premise of allowable error.Further research direction can be listed as follows:combination optimization of access control and resource allocation in access side;optimization of active queue management algorithm by distinguishing the characteristics of flow in backhaul network;optimization of energy division in data compression and research of constant bit rate coding and the multimode data compression algorithm in fronthaul network.