An achievable performance upper bound for discrete multitone equalization

Discrete multitone is a popular discrete Fourier transform based implementation of multicarrier modulation for wireline communications. The paper examines the performance of a complex-tap filter bank structure for channel equalization in discrete multitone modulation systems. The structure under study passes the received signal through a number of logical paths. Each path takes care of one data carrying subcarrier. More specifically, each path cascades a finite impulse response frequency selective equalizer and a Goertzel filter computing a single point discrete Fourier transform. The delay on each logical path is individually optimized for best performance. In the presented simulations testing the achievable bit rate on an asymmetric digital subscriber line transceiver, the filter bank appears to benchmark the bit rate performance among existing discrete multitone equalization methods. An iterative training procedure, which depends on the second-order statistics of the input and output sequences, is proposed to show the achievability of performance.