【摘要】 通常有2类重要的MMSE类盲均衡算法:恒模盲均衡算法和判决引导算法。前者面临收敛速度慢,剩余误差大的缺点;而判决引导均衡算法的收敛速度快,稳态性能好,缺点是启动条件不满足时算法会发散。本论文提出了一种MMSE类盲均衡算法的凸组合策略。通过应用凸组合的基本原理,恒模盲均衡算法可以自适应平滑地切换到判决引导模式。首先,提出了算法的凸组合策略,使得恒模盲均衡算法和判决引导算法之间可以自动软切换。然后描述了一种锚定过程,可以消除稳态时恒模均衡器过高的剩余误差,因此总的算法可以看作是经过了恒模均衡器训练的判决引导均衡。仿真实验证实了新算法的优异性能。更多还原

【Abstract】 There are two kinds of different MMSE blind equalization algorithms:constant modulus(CM)equalization and decision directed algorithm.The former one suffers low convergence rate and high residual error, while the latter always diverges if it is incorrectly started-up.This paper describes a new strategy for the blind equalization so that the CM algorithm can be smoothly switched to the decision-directed (DD) equalization.Firstly,we propose a convex combination approach by running the CM and DD equalization algorithms simultaneously to obtain a smooth switch between them, which is guaranteed by the basic theory of convex combination method. We then describe an"anchoring process"to eliminate the effect from the CM equalization at the steady state to achieve low residual noise. The overall equalization can be regarded as the DD equalization being anchored by the combination approach.Compared with some other methods,our proposed algorithm has the fastest convergence speed and best steady-state performance,which is verified by the numerical simulation.更多还原