On The Interference Alignment Technique With Non-Ideal Channel State Information

After decades of fast development in wireless communication technology, there has been significant progress made in the achieved transmission data rate, spectrum utilization efficiency and other communication metrics. But with the increase in the number of users, interference becomes an important dominating constraint in the achieved performances of wireless communication system. Hence, the reasonable interference suppression technology becomes one of the center research challenges. In2008, V.R.Cadambe et al put forward the interference alignment (IA) technique. By constraining interference while maintaining a reasonable interference-free degree of freedom, IA has received extensive research attention since its presence. The fundamental of IA technique is to suppress the interference within the same signal subspace while leaving the rest subspace for the desired signal by employing the precoding matrix at every individual transmitter. Every receiver can readily recover its desired signal by using the predefined zero-forcing matrix. This thesis starts with the feasibility condition analysis of IA, the partial IA (PIA) technique with interference detection(ID) are utilized jointly to eliminate most of the interferences by using PIA while to suppress the remaining interference by using IC technique. Since IA needs to derive the precoding matrix according to the available channel state information (CSI). the accurate CSI is the pre-requisite to realize IA. But in time-varying channel, it is generally not possible to have the perfect CSI at the transmitter side due to either the limit in feedback bandwidth or the feedback delay. This thesis will lay the emphasis on the IA techniques with non-ideal CSI at the transmitter side caused by either the limit in feedback bandwidth or the feedback delay.Firstly, the thesis reviews the basic principles of IA, iterative IA, alternating minimization IA and Max-SINR IA, The IA feasibility condition analysis without symbol extension for symmetric system shows that, when the user numbers K≤(M+N)/d-1(K is user numbers. M and N are transmitted and received antenna, d is degrees of freedom), interference can be fully aligned at every receiver; on the other hand, for asymmetric system, when interference can not be fully aligned and residual interference exists at the receiver. According to the IA feasibility conditions and interference detection, it is proposed to apply PIA scheme for the interference channel model with path loss to dynamically align inteference at every receivers, while interference dection is utilized to cope with the Residual interference. It is shown through numerical analysis and simulation that PIA and interference detection can achieve larger degree of freedom and better reliability when compared with the random PIAID, iterative IA and Max-SINR IA schemes.Secondly, for non-ideal CSI caused by limited feeback, the interference alignment techniques in a two-cells two-users MIMO-MAC system is investigated in this thesis. The loss in the sum data rate due to the limited feedback and its lower limit are derived to show the difference by using IA with perfect CSI feedback and limited CSI feedback. Our analysis unveils that, given limited CSI feedback, only when the feedback bit width is large enough can the IA outperform the traditional orthogonal TDMA scheme in MIMO-MAC system. In addition, numerical analysis is also presented to confirm that the derived sum rate lower bound can be utilized to well approximate the loss in IA due to the limit in the CSI feedback for MIMO-MAC system. By taking into consideration of the channel spatial correlation and the path loss in dynamic feedback CSI for MIMO interference channel model, the analysis in this thesis shows that better sum data rate can be realized owing to the known apriori spatial correlation and path loss dynamic.Finally, for non-ideal CSI caused by delayed feeback, the zero-forcing beamform (ZF) and MAT IA over block fading model are investigated to show the corresponding degrees of freedom. It is shown that, when transmitted antenna M=2,user numbers K=3,delayed rate0≤γ≤1/3. two degrees of freedom of space-time interference alignment (STIA) can be achieved, which is exactly the achieved degree of freedom by using ZF beamforming over the same block fading channel. After a general review and discussion of the STIA technique over block fading model, a generalized STIA technique is proposed, and it is shown that, given the user numbers K, antenna numbers M=K-1as well as the delayed rate0≤γ≤1/K, the STIA is able to achieve min (K,M)=min(K,K-1)=K-1degrees of freedom over block fading channel, which convinces that the STIA obviously outperforms both ZF-TDMA and ZF-MAT.