I. Introduction
To maximize the data rate of serial communication through a band-limited channel, optimal equalization schemes have been chosen for both the transmitter (TX) and the receiver (RX). A channel that has ≥ 30-dB loss presents severe inter-symbol interferences (ISI) with a significant pre-cursor and a long tail of post-cursors in the received signal, and thus requires a non-trivial equalization scheme, which costs large power consumption and complexity [1], [2]. The pre-cursor must be eliminated using a feed-forward equalizer (FFE) at the TX side, whereas the post-cursors can be compensated for both by FFE at the TX side or by a decision-feedback equalizer (DFE) at the RX side [3]–[6]. However, elimination of the pre-cursor also significantly reduces the main signal due to a large dispersion [7]. To reduce the required channel bandwidth for a given data rate, multilevel signaling with duobinary and PAM4 encodings have been investigated [8]–[12]. Although these encodings reduce the required channel bandwidth by half, they suffer from a small effective signal swing, and from distortion by limited supply headroom [13]. In addition, a large channel loss still necessitates complicated equalization schemes [14]. This brief proposes a channel-included filtering FFE scheme that is suitable for data transmission through a heavily-dispersive lossy channel. The proposed transceiver circuit enables data recovery up to 32-dB loss with only 3-tap FFE at the TX and 1-tap DFE at the RX.