Signal Processing - September 2017 - 142

In this article, we survey the state of the art and research
challenges in the design of signal processing algorithms for the
G.fast system, with a focus on novel research approaches and
design considerations for efficient interference mitigation in
G.fast systems. We also detail relevant very-high-speed DSL
(VDSL) techniques and point out their strengths and limitations for the G.fast system.

The first generation of the G.fast system uses up to 106 MHz,
whereas the next generation goes up to 212 MHz. At this higher bandwidth, G.fast systems suffer from significantly stronger
crosstalk because of the electromagnetic coupling with different lines. With increasing frequency, the crosstalk coupling
between the lines attains the same strength as the direct path
and destroys the diagonal dominance of the channel. This
strong crosstalk poses challenges that cannot be resolved with
Introduction
existing VDSL technology. Hence, the G.fast system requires
DSL has evolved into a viable technology for last-mile access
newer and more powerful techniques for interference mitigain telecommunication networks [1], [2]. This technology levertion, also known as crosstalk cancelation.
ages the existing infrastructure of telephone lines to provide
Signal processing techniques have played an important
affordable broadband services when deployment of optical
role in improving the quality of communications over copper
networks is unfeasible or costly. Since its inception in the
cables and hold the key for future services. In addition to pro1980s, DSL has been considered an interim technology to fill
viding an overview of the current state of the art and research
the gap until the advent of all-optical access networks.
challenges in the design of signal processing algorithms for the
However, we are far away from this horizon. DSL thus
G.fast system, this article focuses on multiuser crosstalk canremains a widely used broadband access technology and will
celation schemes.
play a key role in the convergence of mobile and fixed techMultiuser signal processing, referred to as vectoring by
nologies for next-generation networks.
Ginis and Cioffi in [9], enabled VDSL systems to exceed 100
The upcoming G.fast standard [3], [4] promises to
megabits/s. This multiuser signal processing takes place at a
achieve fiber-like data rates by exploiting the higher bandcentral point that concentrates copper wire pairs from many
width of the telephone lines than previous standards [5],
users. The processing is on both the signals transmitted from
[6]. G.fast is expected to deliver gigabit
the central point to the end users and on
speeds over short loop lengths, as anticithe signal received from the users. Thus,
The fact that DSL systems
pated by Cioffi et al. [7], [8] more than
the vectored system resembles the mulare baseband eliminates
a decade ago. This new standard is very
tiuser multiple-input, multiple-output (MUthe detrimental effect
different from its predecessors in many
MIMO) more than the standard MIMO.
of phase noise, allowing
respects. While fiber is deployed more
However, the vectored DSL system is differand more into the network, rewiring
ent from wireless MIMO in a few important
for a very high spectral
houses with fiber is extremely expensive.
respects. The DSL channel has long channel
efficiency that can reach
The G.fast is a fiber-to-the-distributioncoherence, so CSI can be estimated fairly
15
2 quadrature amplitude
point (FTTdp) technology, which takes
efficiently for multiuser processing. Since
modulation.
fiber to a distribution point (DP) very
users are connected with a fixed modem,
close to the customer premise equipment
channel tracking is also not difficult. More(CPE). The new G.fast standard changed some very fundaover, the telephone channel offers a fairly good channel, and
mental design choices used in VDSL and earlier standards.
the signal-to-noise ratio (SNR) can be as high as 60 dB at low
Most notably, technologies such as time-division duplexfrequencies. The fact that DSL systems are baseband elimiing (TDD) and reverse power feeding (RPF) are used, and
nates the detrimental effect of phase noise, allowing for a very
discontinuous operation is incorporated for energy-effihigh spectral efficiency that can reach 2 15 quadrature amplicient transmissions.
tude modulation (QAM).
The TDD scheme avoids near-end crosstalk (NEXT)
As the G.fast standard uses much higher frequencies
and facilitates asymmetry in downstream and upstream data
than its predecessor VDSL standard, it can no longer rely on
rates more efficiently. It also simplifies channel estimation in
some of the key features of VDSL. One of these features is
downstream transmissions by exploiting channel reciprocity
the well conditioning of the channel matrix. For VDSL frefor channel state information (CSI). The RPF technology simquencies, the DSL channel matrix is column-wise diagonal
plifies the powering of the DP by using the power from the
dominant (CWDD) in the upstream and row-wise diagonal
CPE. This eliminates the need for a power infrastructure at
dominant (RWDD) in the downstream [9]. This characterthe DP and reduces deployment costs. Discontinuous operaistic is very convenient for far-end crosstalk (FEXT) cantion optimizes energy consumption by incorporating lowcelation, and most techniques for multiuser interference
power modes as well as switching off the circuitry at the DP
mitigation in VDSL rely on the diagonal dominance of the
corresponding to the users in offline mode. The G.fast system
channel matrix. This diagonal dominance enables efficient
requires efficient signal processing techniques to harness the
implementation of nonlinear as well as linear crosstalk canbenefits of these features.
celers [9]-[11]. The linear zero-forcing (ZF) precoder in
The G.fast also uses considerably wider bandwidth than
the downstream and the linear ZF canceler in the upstream
the VDSL system, which uses the spectrum up to 30 MHz.
are near optimal for the VDSL channel. These schemes
142

IEEE SIGNAL PROCESSING MAGAZINE

September 2017

Table of Contents for the Digital Edition of Signal Processing - September 2017