Research Of Variable Tap-Length Adaptive Filtering Technology

Single-carrier system is a dominant transmission system in the short wave communications where the Decision Feedback Equalizer (DFE) is wildly used to recover the signal distortion. In practice, the hardware constraints limit the possible length of the time domain equalizer. In addition, the ordinary time-domain equalizer has fixed taps. So it cannot perform equally well in the time varying channels, especially in the severe fading channels (etc. short wave channel). The concept of Variable Tap-Length Equalizer (VLE) attracts researcher's attention after having been proposed. The Variable Tap-length Equalizer has better performance and lower counting complexity. There are four main Variable Tap-Length Algorithms: Segment Filter Algorithm (SF), Gradient Descendent Algorithm (GD), Fractional Tap-Length Algorithm (FT) and Energy Estimation (EE) Algorithm.Among these Algorithms, FT algorithm is the best one, which has fastest convergence rate and most stable steady-state. But FT algorithm also has disadvantages. The performance of FT algorithm deteriorates in the low noise and fast changing systems. To solve this problem, a better FT algorithm was proposed. Compared with the ordinary FT algorithm, the new FT algorithm has faster convergence rate and more stable steady-state than that of the original FT algorithm.The Variable Tap-Length Algorithm mentioned above is based on linear LMS adaptive filter, which can not be directly used on DFE-RLS equalizer. Combining the advantages of FT algorithm and EE algorithm, this paper proposes Fractional EE algorithm which can be used on DFE-RLS equalizer. To evaluate the performance of the VLE, a simplified Watterson shortwave channel model was first constructed. Then the Bit Error Rate (BER) and Computational Complexity are tested. Simulation results show that the DFE-RLS VLE could track the best channel order effectively. Therefore, comparing with the ordinary DFE-RLS equalizer, DFE-RLS VLE could effectively reduce power consumption without harming its performance.