Pulse or digital communications – Equalizers – Automatic
Reexamination Certificate
2000-01-11
2004-01-13
Bocure, Tesfaldet (Department: 2631)
Pulse or digital communications
Equalizers
Automatic
C375S350000
Reexamination Certificate
active
06678319
ABSTRACT:
BACKGROUND OF THE INVENTION
i. Technical Field
This invention pertains to the field of high-speed communications, and more specifically, to signal recovery systems and methods.
ii. Background Art
High-speed communication is a primary focus in the computer and information fields. High-speed communication is desirable in internal networks, external networks, chip-to-chip communications, and any other application in which large amounts of data must be transferred quickly. In a typical local area network, Ethernet connections provide data rates of between 10 Mbps and 100 Mbps. However, this bandwidth is quickly becoming insufficient for modern applications. Data rates in the gigabit range are more optimal for modern applications. Despite the limitations of Ethernet technology, it still provides advantages over competing solutions, including simplicity, low cost, network reliability, availability of management tools, and a high marketplace acceptance. Therefore, a solution to providing higher bandwidth should be preferably compatible with Ethernet technology to be most effective. An IEEE standard has been promulgated for using Ethernet technology for gigabit applications.
FIG. 1
illustrates a simplified block diagram of the functional elements of gigabit Ethernet over copper protocol, as defined in the IEEE standard IEEE Std 802.3ab, approved Jun. 28, 1999. The Media Access Control (MAC) layer
112
arbitrates transmission among all nodes attached to the network. It supports both half and full duplex (transmission/reception) operations. The Gigabit Media Independent Interface (GMII)
108
is a digital interface for carrying unencoded data over separate transmit and receive paths. It connects the MAC
112
to various Gigabit Ethernet physical layer components (such as copper). The 1000 Base-T Encoder/Decoder
104
codes and decodes the signals to be sent and received over the physical layer. Finally, the 1000 Base-T Transceiver
100
contains the physical transmitter and receiver used to transmit and receive the high data rate transmissions. The transceiver
100
is coupled to the physical transmission medium.
FIG. 2
illustrates a block diagram of the IEEE-specified transceiver
100
. A receiver
200
and a transmitter
204
are coupled to a resistive hybrid
208
. The resistive hybrid
208
enables bidirectional transmission over single wire pairs by filtering out the transmit signal from interfering with other signals received by the receiver
200
. The receipt of the transmit signal at the receiver
100
is called near-echo. As there are four pairs of wires for a category-5 cable, four sets of the components illustrated in
FIG. 2
are used.
The most widely deployed cabling system for local area networks is unshielded twisted pair legacy Category 5 copper wiring. Therefore, a receiver enabling high bandwidth transmission preferably accounts for this existing physical infrastructure and its inherent signal impairments. Transmitting a 1000 Mb/s or higher data stream over four pairs of Category 5 unshielded twisted pair cables introduces several channel impairments that were not present or as significant when 10 Mb/s or 100 Mb/s data streams were being transmitted over the Ethernet. These impairments include signal attenuation, i.e., the signal loss due to cabling from transmitter to receiver; echo, which occurs when transmit and receive signals occupy the same wire pair and interfere with each other; noise and intersymbol interference, defined as any unwanted disturbance in the communication channel; and crosstalk, which arises from close coupling of adjacent wire pairs. All of these impairments lead to degradation of the signal quality. Digital signal processing techniques are typically used to address these impairments, including through the use of adaptive filters in a receiver to eliminate intersymbol interference due to attenuation, adaptive filters for echo canceling, and adaptive filters for crosstalk elimination.
FIG. 3
illustrates the IEEE standard implementation of the receiver
200
. A received signal
301
is received over one of the twisted pairs into the resistive hybrid
208
. Next, the analog receive filter
300
receives the analog signal and filters out general noise. Then, the received analog signal is converted into a digital signal through a analog-to-digital converter
305
, through the input of a clock recovery unit
308
that derives a clock signal from the received signal as input from the analog-to-digital converter
305
and adder
316
. The clock recovery unit
308
receives the signal to generate the clock. The clock recovery unit
308
synchronizes the receiver to the carrier frequency. Then, the digital linear feedforward equalizer
312
eliminates any intersymbol interference. The output of the digital linear feedforward equalizer
312
is fed to an adder
316
, which also receives the outputs from four other filters. The other filters provide crosstalk and echo compensation. An echo canceller
320
receives the signal
301
and provides a value that indicates the error introduced from the reflection of the signal transmitted from the transmitter
204
reflected back off the destination. Crosstalk canceller
324
(
1
) receives a signal
302
transmitted across one of the other twisted pair. The crosstalk canceller
324
(
1
) produces a value indicating the error introduced from the coupling of the twisted pair carrying signal
302
to the twisted pair carrying signal
301
. The other crosstalk cancellers
324
(
2
),
324
(
3
) produce values indicating the error introduced from the coupling of the other twisted pairs. The number of taps per channel required for each of these filters have been shown to be between 8 and 16 for the Feed Forward Equalizer, 19 and 14 for the Decision Feedback Equalizer, 70 and 80 for the Crosstalk Cancellers, and 60 and 120 for the Echo Canceller. The adder
316
then provides a composite signal
306
that represents the received signal
301
filtered for the above noise factors. The output
306
of the adder
316
is fed to a Viterbi decoder
328
that uses coding to help recover transmitted symbol in the presence of high noise. The output of the Viterbi decoder
328
is the recovered data
307
. The output
307
is also fed to a decision feedback equalizer
332
that adjusts the adder
316
to compensate for any error detected in transmission.
The IEEE standard does not specify the implementation of the digital filters
312
,
320
,
324
, and
332
. Existing filters require extensive training to learn the correct filter parameters that will overcome the channel impairments. In these filters, a sample of the input signal is used in a training mode at the receiver
200
to determine the characteristics of the channel, and to compensate for those characteristics. These filters then use a variety of digital signal processing techniques to generate the channel characteristics. However, to ensure that the channel characterization remains accurate, the training of the filter must be repeated each time a link is lost, the channel dynamics change, or the noise characteristics vary. Thus, networks that rely on training waste valuable bandwidth, as the time required for training is used at the expense of actual transmission time.
In an attempt to eliminate the requirement of training, other systems use receivers having blind adaptive filters. Blind adaptive filters use only the amplitude of the input signal to recreate channel characteristics.
FIG. 4
illustrates a block diagram of a blind adaptive filter. A transmitter
400
at a remote site transmits a signal s
t
through a channel
404
to an equalizer
408
. Channel
404
adds noise and intersymbol interference to the signal s
t
resulting in a corrupted signal x
t
. The equalizer
408
is designed to recover the input signal s
t
from the signal corrupted by intersymbol interference and noise x
t
. To determine the parameters of the equalizer
408
, an amplitude square extractor
412
is used to provide the square of the amplitude of the received signal x
t
to the bli
Bocure Tesfaldet
Canon Kabushiki Kaisha
Fitzpatrick ,Cella, Harper & Scinto
LandOfFree
Digital signal processing for high-speed communications does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Digital signal processing for high-speed communications, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Digital signal processing for high-speed communications will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3256393