Research On Interference Alignment In Heterogeneous Network

In the Heterogeneous Network(Het Net),the interference problem is particularly serious in the way that macro cells and small cells are overlapped and they use the same spectrum resources to improve spectrum utilization.In the early stage,processing interference as noise,demodulating interference signals or orthogonal multiplexing resources and other technologies,have been difficult to meet the actual demands of interference management.Interference Alignment(IA),as a revolutionary interference elimination technology,provides new ideas for interference processing.Therefore,this paper will use IA to eliminate both intra-user interference and inter-layer interference in the Het Net simultaneously.In this paper,it is firstly introduced about the research background and the current status of IA in the Het Net.And then it is introduced in details about the related concept of Multiple-Input Multiple-Output(MIMO)and Het Net,as well as the basic idea and implementation methods of IA.Secondly,an Optimized Generalized Eigenvalue Decomposition Interference Alignment algorithm(OGEVD-IA)is proposed to solve the interference problem in the uplink two-layer Het Net.The proposed algorithm can also eliminate intra-user interference and inter-layer interference effectively under the condition that the channel state information at transmitter is imperfect.The simulation results show that compared with Max-SINR,OGEVD-IA not only improves the SNR by 0.6bps/Hz at 20d B,but also improves the SNR by 2d B at10^-3 BER.Finally,the Graded Interference Alignment(G-IA)algorithm is proposed to solve the interference problem in the downlink 3-layer Het Net.In the G-IA algorithm,the Block Diagonalization(BD)algorithm is firstly used to eliminate the interference between users of each layer,and then different weighting factors are used to deal with the different layers to eliminate inter-layer interference.The simulation results show that the G-IA algorithm improves the sum-rate by 0.5bps/Hz compared with the Max-SINR algorithm at 20d B SNR.Additionally,with the increasing of SNR,this performance difference becomes more obvious.