[1] Huang Yongming, Yang Luxi, Bengtsson M. A limited feedback joint precoding for amplify-and-forward relaying[J]. IEEE Trans on Signal Processing, 2010, 58(3): 1347-1357. [2] Xu Wei, Dong Xiaodai, Lu Wusheng. MIMO relaying broadcast channels with linear precoding and quantized channel state information feedback[J]. IEEE Trans on Signal Processing, 2010,58(10): 5223-5245. [3] Jinal N. MIMO broadcast channels with finite-rate feedback[J]. IEEE Trans on Information Theory, 2006, 56(11): 5045-5060. [4] 解芳,袁超伟,程铁铮,等. MIMO广播系统中基于有限反馈的多用户选择新算法[J]. 电子与信息学报, 2010, 32(11) : 2785-2789. Xie Fang, Yuan Chaowei, Cheng Tiezheng, et al. A new multi-user selection algorithm in MIMO broadcast system with limited feedback[J]. Journal of Electronics and Information Technology, 2010, 32(11) : 2785-2789. [5] Ravindran N, Jindal N. Limited feedback-based block diagonalization for the MIMO broadcast channel[J]. IEEE Trans on Selected Areas in Communications, 2008, 26(8): 1473-1482. [6] Dabbagh A D,Love D J. Mutiple antenna MMSE based downlink precoding with quantized feedback or channel mismatch[J]. IEEE Trans on Communications, 2008, 56(11) :1859-1868. [7] Gupta A, Nagar D. Matrix variate distributions[M]. London:Chapman, 2000:165-170. [8] Skovgaard C S, Rajiv A, Elisabeth D C, et al. Weighted sum-rate maximization using weighted MMSE for MIMO-BC beamforming design//ICC 2009. Dresden: IEEE Press, 2009: 1-6. [9] Muirhead R J.Aspects of multivariate statistical theory[M].New York:Wiley-Interscience,1982:65-83.附录1由文献 [7] 可知,Xk和Yk是相互独立的,Sk与Zk是相互独立的.由文献 [9] 得到Z</em>HkZk服从Beta分布,即Z</em>HkZk~Beta(N,M-N).根据Beta分布的性质 [7] ,YHkYk~Beta(M-N,N),且由文献 [7] 可知Zk的对角元素z2ii~Beta,则1) E[XkYkΔ1/2k]=E[XkYk]E[Δ1/2k]αIN.2) E[Y</em>HkX</em>HkΔXkYk]=ME[XkYkX</em>HkY</em>Hk]=ME[YkXkX</em>HkY</em>Hk]=ME[YkY</em>Hk]=(M-N)IN,其中由Beta 分布性质 [7] 可知E[YkY</em>Hk]=IN.3) E[ZkΔ1/2kZ</em>Hk]=E[Zk] E[Δ1/2k] E[Z</em>Hk]=ME[ZkZ</em>Hk]=IN,其中E[tr[ZkZ</em>Hk]]=D,则E[ZkZ</em>Hk]=IN.4) 为了证明得到最后结果,设W和W⊥组成正交的空间,则Sk可以表示为Sk= W⊥Ok,其中Ok为酉矩阵,平均分布在M-N维空间中,则E[SkS</em>Hk]=E[W⊥OkO</em>HkW⊥H]= E[W⊥W⊥HOkO</em>Hk]=×E[W⊥W⊥H]=(IM-k</em>Hk). |