摘要
本文推导了基于梯度算法的最小均方误差二维判决反馈均衡器的时变步长选择算式，它在最速下降均方误差意义上来讲是最优的。与固定步长的ＬＭＳ—ＤＦＥ相比具有收敛快，在时变信道下也具有极强的跟踪能力和好的误码性能等优点，而计算复杂性却增加很小。计算机模拟结果证实了上述结论。
his paper derives a formula of time-variant convergence step for two -dimension adaptive decision feedback equaliger(DFE),which uses the gradient type least mean square (LMS) algorthm,this time-variant convergence step LMS-DFE(TLMS-DFE)is computationally more efficient than the fixed iteration step. its performances have beem inversitaged on both time-invariant and time-variant channels through computer simulation. The simulation results show that the TLMS-DFE has the advantage of fast initial convergence rate,and has good capablity of tracking rapid time -variant channels. However it has the approximately same computational complexity as the LMS-DFE.
关键词
反馈均衡器
时变步长
数字通信
two-dimension adaptiou decision-feedback equalizer
