ATSC RF, Modulation, and Transmission

advertisement
ATSC RF, Modulation, and Transmission
WAYNE BRETL, SENIOR MEMBER, IEEE, WILLIAM R. MEINTEL, GARY SGRIGNOLI,
XIANBIN WANG, MEMBER, IEEE, S. MERRILL WEISS, AND KHALIL SALEHIAN
Invited Paper
The developmental aspects and technical characteristics of the
ATSC RF transmission standard (“8-VSB”) are presented. An exposition is given of the planning and allocation methods that were
developed, which are generally applicable to the introduction of a
simulcast DTV service independent of the type of modulation used.
Additional modulation enhancements (E-VSB) are explained. Techniques for implementation of distributed networks of on-channel
transmitters are introduced along with references to some specific
applications of these techniques.
Keywords—Advanced Television Systems Committee (ATSC),
digital television broadcasting, distributed transmission networks,
spectrum planning.
I. INTRODUCTION
On 13 February 1987, a petition was filed with the U.S.
Federal Communications Commission (FCC) requesting
that the FCC initiate an inquiry to explore the impact of
advanced technologies on existing television broadcasting
and the FCC’s spectrum allocation policies relating to television. In response to that petition the FCC, on July 16,
1987, adopted a Notice of Inquiry (NOI) covering a broad
range of issues relating to over-the-air television service [1].
At the time of the NOI and in the three years to follow,
numerous ideas to improve broadcast television were put
forward. The proposals ranged from modest to major improvements and from fully backward compatible systems to
completely incompatible new designs. Some proposed the
use of new spectrum while others focused on using the spectrum already allocated for broadcast use.
Manuscript received June 28, 2005; revised September 19, 2005.
W. Bretl is with the Zenith Electronics Corp., Lincolnshire, IL 60069
USA (e-mail: wayne.bretl@zenith.com).
W. R. Meintel is at P.O. Box 907, Warrenton, VA 20188 USA.
G. Sgrignoli is at 1139 Juniper Lane, Mount Prospect, IL 60056 USA.
X. Wang and K. Salehian are with the Communications Research Centre,
Ottawa, ON K2H 8S2, Canada.
S. M. Weiss is with the Merrill Weiss Group LLC, Metuchen, NJ
08840–1242 USA.
Digital Object Identifier 10.1109/JPROC.2005.861018
In 1990 the FCC made the decision that the implementation of an advanced television system (ATV) in the United
States would be accomplished within the existing television
bands and that ATV channels would have the same 6-MHz
bandwidth as the existing analog National Television Systems Committee (NTSC) channels. Unlike the introduction
of color television in the 1950s, it was decided that it was not
practical to design an ATV system that allows for transmission of analog and ATV on the same channel. In view of this
decision, it became necessary to find an additional channel
for each existing NTSC station so they would be able to begin
transmission of ATV while continuing to provide NTSC service during a transition period.
This paper details the development of planning tools for
implementation of a digital DTV system that is simulcast
with existing analog transmissions. It then presents the
features and characteristics of the ATSC 8-level vestigal
sideband (8-VSB) transmission standard, and the most
recent developments regarding that standard, including
enhanced compatible modulation methods (E-VSB) and the
use and implementation of single-channel multiple transmitter networks.
II. CHANNEL ALLOTMENT PLANNING
The goal of the allotment planning was to providing a
second channel for each of the approximately 1700 full service television stations within the United States. However,
at the time this work was being carried out, there were still
several competing ATV system proposals, and each system
had a different set of allotment planning factors that had to
be accommodated by the computer modeling.
The first efforts to determine the feasibility of developing a
plan to implement ATV were carried out by Robert Eckert at
the FCC. Soon thereafter a group of broadcast organizations
undertook the funding of a project, coordinated by the Association for Maximum Service Television (MSTV), to develop
computer modeling to assess the availability of spectrum for
ATV.
The initial work of this project was an expansion of the
FCC efforts whereby channel assignments were made on the
0018-9219/$20.00 © 2006 IEEE
44
PROCEEDINGS OF THE IEEE, VOL. 94, NO. 1, JANUARY 2006
basis of desired separation distance between stations. Improvements to the FCC’s Assignment Model were made to
allow the study of various separation criteria as well as the
consideration of adjacent and UHF taboo1 channels, plus
special criteria for colocated stations. Although this method
created plans and allowed the determination of the number
of existing analog stations that could be accommodated with
a second ATV channel, it did not provide any information
concerning expected coverage and interference.
Because a nationwide plan for the implementation of ATV
would have such a profound effect on the broadcast industry
well into the next century, it was imperative that its design
employ the best possible methods. In view of that requirement, efforts were undertaken to improve on the Assignment Model software to address the issues of coverage and
interference.
1) Model Development: To determine the spectrum
requirements for implementation of the ATV system, two
computer models were developed and integrated. The
Assignment Model generated spectrum-efficient channel
assignment plans based on minimum separation distances;
while the Coverage and Interference Model (CIM) evaluated
the performance of plans generated by the Assignment
Model. The evaluation provided the expected coverage and
service lost due to interference based on a set of user supplied system parameters. In addition, the CIM also had the
capability to perform channel substitutions when its analysis
indicates that such a change would yield an improved plan.
2) Assignment Model: The Assignment Model used
minimum separation distances to determine the number
of existing stations that could be accommodated with an
additional ATV channel under different cochannel, adjacent
channel, and taboo channel distance separation criteria.
Specifically, the model used a heuristic approach to determine the “best” ATV accommodation statistics for a given
set of input conditions. This was accomplished by first ordering the existing analog stations for a given area according
to the apparent difficulty of finding a channel for them
and then using a mathematical optimization method to find
the largest number of stations that can be accommodated
within that area. The output of the model was an assignment
table that paired existing NTSC stations with specific ATV
channel assignments.
Two basic modules are used within the Assignment Model.
The first module, the constraint generator, determines for
each existing NTSC station all possible ATV channels that
could be assigned to that station based on the user supplied
separation criteria. The second module, the optimizer/evaluator, used a number of mathematical algorithms to iteratively
search for the lowest number of ATV channels that could be
assigned to accommodate the entire set of NTSC stations.
Two different techniques were used to investigate the ATV
assignment problem. The first technique employed an al1The channel relationships commonly referred to as the UHF taboo channels include the channels 2, 3, 4, 5, 7, 8, 14, and 15 with respect to
the desired channel. Although they are usually discussed separately, the first
adjacent channels are also technically referenced in the FCC’s Rules as part
of this group.
6
+
+
BRETL et al.: ATSC RF, MODULATION, AND TRANSMISSION
gorithm developed by Frank Box [2]. This technique is a
heuristic assignment approach that operates on the principle
that stations should be assigned channels in descending order
of assignment difficulty. Through an iterative process, stations that repeatedly fail to be assigned channels, rise toward
the top of the list of requirements for channels and are accommodated first. The successive reordering of stations by order
of difficulty tends to improve accommodation statistics of a
resulting solution or plan.
The second technique, developed by Robert Eckert at
the FCC, uses a general algorithm, known as simulated
annealing [3]. This method attempts to solve the assignment
problem by starting with a random solution and repeatedly
proposing small changes to the solution, then replacing the
solution each time an improved accommodation is obtained.
Both of these techniques were used to investigate the assignment question, and Eckert indicated that the simulated
annealing approach yielded somewhat better results.
A. Coverage and Interference Model
Initially, software was developed that provided coverage
and interference area determinations for a single station with
an option to graphically display the results. Computations
were made using the FCC R-66022 propagation curves
and user-specified station data. From this basic concept, a
highly sophisticated computer model evolved. The enhanced
model provided the capability to analyze nationwide ATV
assignment plans, with flexibility to vary station and system
parameters, select from any of three propagation models,
and furnish statistics on population density within coverage
and interference areas for individual stations and for the
entire country. The model also was improved to employ
actual station data from the FCC’s Television Engineering
and Directional Antenna databases as well as the ability to
use three-arc-second terrain data as needed for propagation
modeling. In addition to the many computational features
provided, the software also incorporated capabilities for
optimization of ATV assignment plans.
1) Coverage and Interference Computation Procedure
(Initial Model Options): The extent of the noise-limited
service contour of a station is determined using the FCC
propagation curves. The F(50,50) curves are used when
computing NTSC service, and at the user’s option either
F(50,50) or F(50,90) curves can be used for ATV computations. (The F(50,90) curves are preferable for digital
ATV planning because of the all-or-nothing nature of digital
reception compared to analog’s “graceful degradation.”)
The desired value of the noise limited contour (specified
2The FCC R-6602 propagation curves are contained in the FCC Rules
Section 73.699. They are a family of curves that provide estimates of field
strength exceeded for either 50% of the time or 10% of the time at 50% of
the potential receiver locations. The curves are based on a receive antenna
height of 9 m and provide estimates for various average heights above
average terrain (height above the average terrain (HAAT) between 3.2 and
16.1 km from the transmit location) for the transmit antenna. They are
commonly referred to as the F(50,50) or F(50,10) curves. The F(50,90)
curves used to predict field strengths for 50% of the potential receiver
locations 90% of the time can be derived by the following formula: F(50,90)
F(50,50) [F(50,10) F(50,50)].
=
0
0
45
in decibels above 1.0 V/m, dB ) is a user-specified value
with different values permitted for each system (NTSC or
ATV) and TV band.3 If the user selects the FCC curves
as the propagation model, then it is assumed that field
strength values will equal or exceed the noise limited value
everywhere within the computed contour. However, if either
of the optional propagation models (Longley–Rice [4] or
TIREM [5]) is selected, then a further analysis is made to
predict within the contour where noise-limited service will
exist. This is required since predicted signal level using these
alternative propagation models takes into consideration the
terrain features along the entire path between the transmitter
and the point being evaluated, whereas use of the FCC
propagation curves assumes an average height above the
surrounding terrain in all directions (see note 5).
Included in the computations are transmit power and antenna height above the surrounding terrain, and at the user’s
option, the specific horizontal4 antenna patterns associated
with each station. Power and height are either that found in
the FCC data base or set to a fixed value for each band and
system. When analyzing ATV assignments contained in an
assignment plan, the user can optionally set the power and
height or have the station power computed so that the ATV
noise limited service contour will match that of its paired
NTSC station, using the paired station’s parameters.
Interference calculations are made using the same station
parameters and propagation model criteria used to determine
noise-limited service. If the FCC curves are selected as the
propagation model then the F(50/10) curves are used to compute interfering signal levels. If either the Longley–Rice or
the TIREM model is selected then the user is requested to
specify the desired location and time probabilities, and in the
case of the Longley–Rice model, the level of confidence. The
interference computations also include a standard receive antenna pattern that is assumed to have its main lobe pointed in
the direction of the desired station.
When determining the existence of interference, the user
also can specify the channel relationships to be considered.
Relationships that can be included are cochannel, upper
and/or lower adjacent channels and the UHF taboo channels.
Different channel relationship considerations can be specified for NTSC-to-NTSC, NTSC-to-ATV, ATV-to-NTSC,
and ATV-to-ATV interference computations. Likewise, the
user can specify the acceptable ratio of desired-to-undesired
(D/U) field strength levels for each channel relationship
that is to be considered when predicting the existence of
interference.
In order to evaluate the impact of interference from
individual stations as well as overall interference, and to
assess the amount of new interference caused to NTSC
stations by the introduction of ATV service, a service matrix
concept is employed. This concept divides the area within
3In the United States three bands are allocated for television broadcasting:
low VHF (54–72 MHz and 76–88 MHz), high VHF (174–216 MHz), and
UHF (470–806 MHz).
4Since no antenna vertical plane data is contained in the FCC’s database,
a set of typical standard vertical antenna patterns was developed for each
type station and band.
46
the service contour of a station into a matrix of cells. A
determination is then made of the cells within the service
area where interference is present. The total area within the
contour receiving interference is then determined by adding
up the area of the cells where interference is found to exist.
By overlaying the matrices that define the areas of interference caused by individual stations it is possible to determine
the overall area of interference. Likewise, by determining
the difference between a matrix of the interference caused by
other NTSC stations and a matrix of ATV-only interference
to an NTSC station it is possible to determine the amount of
new interference that would be caused by the introduction
of ATV stations. Similarly, it is possible to show the extent
of interference that will be received by an ATV station when
NTSC stations are no longer operating.
The same matrix concept is used to determine noise limited service when either the Longley–Rice or TIREM propagation model is selected. The area of the cells where the
computed signal level is below the noise limited threshold is
subtracted from the total area within the noise limited contour to determine noise limited service area.
2) Determination of Population Densities: To be able
to ascertain the population within service and interference
areas, a database of the 1990 U.S. census at the block level
was established. This database contains the population and
reference coordinates of each census block in the United
States. After the service contour of a station is established,
the population within each matrix cell within the contour
is computed by determining the total of the population of
the census blocks that fall within the cell. Once this population matrix has been established, it is then used in the
same manner as the area matrix to compute the population
statistics.
3) Optimization Software: In addition to providing the
capability to determine coverage and interference statistics
for an ATV assignment plan, the model also has two different options that attempt to optimize channel assignments
by matching ATV interference areas with interference areas
of the paired NTSC station. The first option attempts to determine the “best” assignment scheme for stations located at
a common site. A common site can be defined as a single
location or locations that are within a specified distance of
each other. The second option attempts to improve the plan
by substituting other channels when such channels are determined to provide improved interference matching.
The first option analyzes all channels assigned at a colocated site for each of the stations at the site and creates a matrix of the mismatch between the existing NTSC interference
area and ATV interference area. The reason to analyze each
channel for each station is that this option assumes that the
ATV assignment service contours are to duplicate those of
the paired NTSC stations; therefore, the coverage and interference of each ATV station will depend on that of the paired
NTSC facility. After all the stations and channels have been
examined, the matrix is then analyzed. The NTSC station
with the poorest overall interference area match (including
all channels) is assigned the ATV channel that has the best
match. That station and channel are then removed from the
PROCEEDINGS OF THE IEEE, VOL. 94, NO. 1, JANUARY 2006
matrix and the process is repeated until all NTSC stations
have been assigned an ATV channel. This system attempts
to insure that matching is evenly distributed among the stations at the common site.
The second optimization option analyzes other available
channels in an attempt to find an ATV channel that either
does not receive any interference or whose interference area
better matches that of its paired NTSC station.
When an ATV assignment plan is created based on specified separation distances, many of the locations may still have
other channels that could have been assigned. This option
first analyzes the planned ATV channel assignment to determine the amount of interference it receives from other ATV
stations and existing NTSC stations. If no interference is received, then no further analysis of the ATV assignment is
made. However, if interference is found, then the area where
the interference occurs is compared to the interference area
of the paired NTSC station. If the interference areas match
exactly (no interference is caused to the ATV station outside
the area where the paired NTSC station received interference), no further analysis is performed; otherwise, each of
the other available channels is analyzed searching for either
a channel that receives no interference or one where the interference area better matches that of the paired NTSC station.
If an improvement is found, that channel is substituted for
the channel originally assigned. The lists of additional channels available at other sites not yet analyzed are then checked
to determine if this assignment would conflict with them. If
a conflict is found, the list for the other sites is modified to
eliminate the conflict. Conflicts are determined on the basis
of the spacing requirements used to create the assignment
plan. It should be noted that, before the available channels are
analyzed by the model, the channels are ranked by distance
to the nearest cochannel NTSC station or ATV assignment
and those with the greatest distance separation are analyzed
first.
4) Modifications to the Initial Model Options: In addition
to the originally planned model options discussed above, it
became necessary to create several others to study the effect
of certain planning concepts not originally anticipated.
Ultimately it was decided that the methodology based on
the Longley–Rice propagation model was the best approach.
This decision was based on the capability to take into consideration the specific terrain characteristics of each path,
which was expected to provide a much more realistic prediction of service and interference than a methodology that only
considered the terrain near the transmitter (FCC propagation
curve model). A comparison of the results produced by
the Longley–Rice model to that of a specific version of the
TIREM model indicated that the overall differences were
not significant. However, due to the fact that at the time
there were numerous versions of the TIREM model, and
only one documented and widely accepted version of the
Longley–Rice model, the Longley–Rice model was adopted.
Although the CIM model, as discussed above, was originally developed by the broadcast industry, it was eventually adopted by the FCC with some minor variation in the
BRETL et al.: ATSC RF, MODULATION, AND TRANSMISSION
methodology used to define the cells used for the coverage
and interference matrices.
The industry model used wedge shaped cells centered on
evenly spaced radials whereas the FCC model used essentially square cells. The FCC used this concept to initially
allow the use of the same matrix grid for several stations
within a market. However, the FCC later abandoned that concept in favor of using a specific matrix for each station based
on the station’s location. The only other deviation was the
evaluation point within each cell. In the FCC model, the evaluation point was determined by the centroid of the population
within the cell rather than the center of the cell as used by the
industry model.
Once developed, the CIM was used both to analyze the
efficiencies of various channel allotment plans and to refine
the plans, and played a major part in the final allotment plan
adopted by the FCC [6], [7]. In addition, the CIM was used to
compare the competing system proposals as well as to assist
the Grand Alliance later in their efforts to achieve consensus
on a single method for the transmission of ATV. (The digital HDTV Grand Alliance was a consortium of all remaning
system proponents formed toward the end of the selection
process.)
After adoption of the final transition allotment plan the
FCC adopted the CIM methodology to process the applications for the newly allotted channels. The CIM has become
known as the FCC OET-69 methodology, which is detailed
in FCC OET Bulletin 69 [8]. This bulletin provides the final
set of planning factors that are to be used to analyze coverage
and interference in the United States. This methodology
was necessitated because the FCC has permitted individual
stations to deviate from the parameters in the transition
allotment plan, including changing channels and increasing
power, provided that any new predicted interference is
less than a specific percentage referred to as de minimus
interference.
B. Longley–Rice Propagation Model
The official name of the Longley–Rice model is the ITS
Irregular Terrain Model. The model computes path loss adjustment to free space signal propagation and is a generalpurpose model covering 20 MHz to 20 GHz. The model
was developed in the 1960s and is semiempirical, using theoretical treatment of reflection, refraction, diffraction, and
scatter with the theory adjusted to fit measured data. It is not
intended to predict short-term variations, as it is based on
yearly medians. The data used to develop the model came
from VHF land mobile service measurement in Colorado
and northern Ohio and from the VHF and UHF television
measurements from the Television Allocations Study Organization (TASO) and also includes some climatic data from
CCIR.
C. Final FCC Transition Planning Criteria
The final ATV allotment plan adopted by the FCC provided a companion channel for each authorized full-service
television within the United States. The facility authorized
47
for the companion channel provided an ATV noise-limited
contour5 that replicated the Grade B contour6 of the station’s
paired analog channel. However, if the ATV facility required
to replicate the analog contour was below the minimum
power level (1.0 kW for low VHF, 3.2 kW for high VHF,
and 50.0 kW for UHF) the power was set to the minimum
value. Likewise if the power on a UHF channel exceeded
1000 kW, the power was set to that level. It is also noted that
a dipole factor adjustment7 was applied to the normal Grade
B or ATV noise limited contour values when determining
the replication that involved a UHF channel. The other
parameters used in developing the final allotment plan can
be found in the FCC OET Bulletin 69 discussed above.
Fig. 1. VSB and NTSC spectral comparison in 6 MHz.
D. Spectrum Issues Then and Now
As noted previously, the initial goal was to provide each
existing full service television station with a second channel
for ATV transition operation. However, no consideration was
given to either protecting the more than 7000 existing low
power television stations, booster stations, and translators
(collectively referred to as LPTV) or providing them with a
transition companion channel.
In addition, midway through the process it was decided
by the FCC, acting on legislation passed by the U.S. Congress, that the number of television channels at the end of
the transition would be reduced from 67 to 49.8 Since the
posttransition television band would include only channels
2–51 instead of 2–69, the final planning process attempted to
avoid the use of channels above 51. This constraint made it
even more important to use the techniques of the CIM to develop an efficient plan that avoided unnecessary interference
to the greatest extent possible. In the end it was not possible
to totally avoid use of the channels above 51.
At the time of this writing (August 2005), the FCC was
in the process of developing a final posttransition DTV9
allotment plan and is making extensive use of the CIM to
create that plan. Once again, the need for the best planning
methodology has taken on new importance in that it has been
determined that low VHF channels (2–6) may not be suitable
for DTV operation in many areas due to excessive manmade
noise. Furthermore, since the frequencies above channel
51 will be used for nonbroadcast purposes, broadcasters
are likely to try to avoid use of channel 51 due to potential
interference issues that could arise both from and to those
other services. In the end, it is expected that use of some
5ATV (DTV) noise limited contours are 28 dB for low VHF, 36 dB for
high VHF, and 41 dB for UHF, and are computed using the FCC F(50,90)
propagation curves.
6Analog Grade B contours are 47 dB for low VHF, 56 dB for high VHF,
and 64 dB for UHF, and are computed using the FCC F(50,50) propagation
curves.
7The dipole factor adjusted value is computed by the following formula:
64–20 log[615/(channel midfrequency in megahertz)] for analog and 41–20
log[615/(channel midfrequency in megahertz)] for ATV (DTV).
8The number of available television channels is reduced by one because
channel 37 (608–614 MHz) is reserved exclusively for the radio astronomy
service.
9During the course of this process, ATV has become DTV. Initially all the
proposed ATV systems were analog, but since the eventual system is digital,
it was appropriate to refer to the new system as DTV.
48
low VHF channels as well as channel 51 will be required to
provide a channel to every station in the congested areas.
The FCC has also adopted regulations that permit LPTV
stations an opportunity to convert to digital (flash cut) or to
obtain a companion digital channel and adopted use of the
CIM methodology to process those requests.
III. THE ATSC VSB TRANSMISSION SYSTEM
An in-depth description of the digital VSB transmission
system has been published in ATSC documents [9], [10]. The
ATSC-standardized version was adopted in the United States
by a series of FCC “Report and Order” documents and published in the FCC Rules [11]–[14]. The following is a brief
overview of the digital VSB transmission system and its performance.
A. ATSC VSB Transmission Description
The VSB and NTSC spectra in a 6-MHz channel are
compared in Fig. 1. The analog NTSC vestigial-sideband
spectrum has the familiar three carriers (visual, chroma,
and aural) that carry most of the energy in the channel. Its
power is described in terms of the peak envelope power of
the large, constant-amplitude horizontal and vertical sync
pulses. Compared to digital modulation, the analog television signal is not efficient in terms of either bandwidth or
power.
In contrast, the digital VSB spectrum is flat throughout
most of the channel due to the noise-like attributes of randomized data. Its power is described as the average power
in the channel bandwidth. Its peaks can only be described in
terms of a statistical cumulative distribution function [15],
with the RF envelope peaks remaining 99.9% of the time
within 6.3 dB of the total in-band average power. Two steep
transition regions (each 620 kHz wide) exist at each end of
the band. The half-power frequencies of the VSB signal are
5.381 MHz apart, and represent both the Nyquist bandwidth
(smallest ideal bandwidth that can sustain the desired ATSC
symbol rate) and the equivalent noise bandwidth. Therefore,
the 6-MHz channel uses only 11.5% excess RF channel
bandwidth, making it a very efficient system in terms of
bandwidth. A low-level constant RF pilot carrier (which
adds only 0.3 dB to the total average power) is added to the
noise-like data signal at the lower band edge. (The small
PROCEEDINGS OF THE IEEE, VOL. 94, NO. 1, JANUARY 2006
Fig. 2. Data segment and segment sync definition.
Fig. 3. Frame sync definition.
in-phase pilot is created by adding a dc component equal
to 1.25 constellation units to the eight equi-probable levels
prior to modulation.) These attributes make the digital VSB
signal very efficient in terms of both power and bandwidth.
The baseband data-segment format for terrestrial broadcasts is illustrated in Fig. 2 for the trellis-coded 8-VSB signal.
Note that there are eight discrete equi-probable data levels
except for the four-symbol binary data segment syncs that
delineate each 832-symbol data segment. Therefore, each of
the 828 data payload symbols carries 3 b of information at
a 10.762-MHz symbol rate. However, only two of these bits
are payload data; the third bit is a parity bit resulting from
trellis-coded modulation.
The syncs are not trellis-coded. These segment syncs take
that occurs at the
the place of the MPEG sync byte
beginning of every fixed-length 188-byte MPEG packet.
Thus, the swap of MPEG-sync for segment sync causes no
loss of data throughput, since the MPEG-sync is reinserted
at the VSB demodulator’s output. In the 8-VSB system,
each data segment carries one 188-B MPEG packet plus
20 B of Reed–Solomon (RS) error correction parity for a
total of 208 B/segment.
BRETL et al.: ATSC RF, MODULATION, AND TRANSMISSION
Fig. 3 depicts a DTV binary data frame sync, which is one
segment long (832 symbols) and repeats every 313 segments.
The data efficiency is reduced by only 0.32% (1/313) due to
the insertion of data frame syncs. The data frame allows data
frame synchronization in the receiver. The same binary frame
sync (particularly the 511-symbol and three 63-symbol PN
sequences) can also be used as a known reference-training
signal for the receiver equalizer, which allows initializing
the equalizer as well as stable operation regardless of data
eye closure. The frame sync can also act as a means of determining received RF signal conditions (such as S/N ratio).
The middle 63 PN sequence alternates from one frame to the
next, providing two unique frame syncs.
The VSB transmission system was developed including
five modes (2, 4, 8, 8-trellis, and 16), all of which appear
in ITU-T Recommendations J.83 and J.84 [16], [17] These
modes provide a tradeoff between data rate and robustness
of reception by means of different numbers of constellation levels and/or the use of trellis coding. The 24-symbol
VSB mode section in the binary frame sync reliably informs receivers which of the modes is being transmitted.
The Grand Alliance and the Advisory Committee on Ad49
Fig. 4.
8-VSB transmitter block diagram.
Fig. 5.
8-VSB receiver block diagram.
vanced Television Service (ACATS) selected trellis-coded
8-VSB (19.39-Mb/s payload) for terrestrial broadcast and
nontrellis-coded 16-VSB (38.78-Mb/s payload) for cable
transmission. Note that the symbol rate is 10.762 MHz for
all of the various VSB modes, with a varying number of
bits/symbol determining the data rate for each mode to trade
off against reception performance.
The trellis-coded 8-VSB-transmitter block diagram is
shown in Fig. 4. The transmitter, which includes the channel
encoder and exciter, receives the incoming data packets
(188 B/packet of interspersed video, audio, and ancillary
data), and thoroughly randomizes the data so that the transmitted signal has a flat, noise-like spectrum. Random data
is important for all the receiver recovery loops to work optimally, and minimizes interference into analog NTSC. The
Reed–Solomon encoding, known for its good burst noise
correction capability and data overhead efficiency, adds
the 20 parity bytes to the end of each MPEG data packet
before data bytes are convolutionally interleaved (spread
out) over 52 data segments, to a depth of 1/6 of a data frame
(4 ms). Segment and frame syncs are not interleaved. Data
byte interleaving helps protect against the effects of burst
noise/interference that occurs during transmission. In the
trellis-coded 8-VSB terrestrial system, the trellis encoder
adds additional redundancy to the signal in the form of more
(than four) data levels, creating the multilevel (eight-level)
data symbols for transmission. The added redundancy allows
further error correction in the receiver through the cascaded
forward error correction made up of Reed–Solomon block
coding and convolutionally coded trellis-coded modulation
(2/3 rate, four-state Ungerboeck code). In 16-VSB cable
applications, there is no trellis encoding, and a mapper
creates the multilevel (16-level) data symbols instead of a
trellis-coder.
50
The segment and frame syncs are then multiplexed with
the multilevel data symbols before the dc offset is added for
creation of the low-level, in-phase pilot. The VSB modulator
provides a (root-raised cosine) filtered IF signal (typically
centered on 44 MHz in the United States), with most of one
sideband removed. Finally, the RF upconverter accurately
translates the IF signal to the desired RF channel with as
little phase noise as possible before amplification by a highpower amplifier (HPA) and emission mask filtering that significantly reduces any adjacent channel energy splatter due to
third-order and fifth-order nonlinear intermodulation. The required adjacent channel emission mask details can be found
in FCC documents [18] and other documents [19]–[21]. An
optional feedback signal from the emission mask filter output
can be fed back through a processor to precorrect the transmitter for any linear distortion caused by the mask filter (e.g.,
nonflat amplitude or group delay response) or any nonlinear
distortion caused by the HPA, such as amplitude clipping or
incidental carrier phase modulation.
The VSB receiver block diagram is illustrated in Fig. 5.
The DTV signal, using the existing 6-MHz RF channel
allocations, is converted by the VSB receiver’s tuner to
the IF frequency (typically 44 MHz) prior to channel decoding. After appropriate IF filtering (root-raised cosine)
and automatic gain control (AGC), the pilot signal can
be used to synchronously detect the VSB signal while simultaneously removing FM sidebands and low-frequency
phase noise inherent in low-cost consumer tuners. The clock
synchronizer recovers the 10.762-MHz symbol clock from
the received signal as well as synchronizes the various
following loops (interleaver, trellis and Reed–Solomon
decoders, and randomizer) so that further VSB processing
can be accomplished. The equalizer is the workhorse of
the DTV receiver removing any linear distortion due to
PROCEEDINGS OF THE IEEE, VOL. 94, NO. 1, JANUARY 2006
Table 1
VSB Transmission System Performance Parameters
Fig. 6. Probability of error versus C/N for 8-VSB and 16-VSB systems.
multipath propagation or imperfect filtering. Any remaining
high-frequency tuner phase noise is removed by the phase
tracker. The remaining data processing circuits perform
the concatenated forward error correction (short four-state
trellis decoding with minimal error propagation followed by
Reed–Solomon decoder that can correct up to ten
a
byte errors/MPEG packet). A convolutional deinterleaver
reassembles the interleaved (dispersed) data bytes while dispersing (spreading out) contiguous burst errors to maximize
Reed–Solomon error correction. Finally, the derandomizer
recreates the original data bytes for the transport multiplexer
to deliver to the video and audio decoders.
The 16-VSB-cable mode is very similar to the terrestrial
mode except there is no trellis-coded modulation, and the
data rate is doubled (using the same symbol rate) by sending
four data bits per symbol rather than two data bits per symbol.
Further details can be found in [22].
B. ATSC VSB System Performance and Compliance
VSB performance has been extensively evaluated in simulation and measured in the laboratory [22], [23]. Table 1 conBRETL et al.: ATSC RF, MODULATION, AND TRANSMISSION
tains the primary system performance parameters and their
values. Note the similarities (common parameters) and differences among the various VSB modes as lower data rates
achieve more robust DTV reception.
Fig. 6 is an illustration of the probability of error versus
signal-to-noise ratio for both the 8-VSB terrestrial and
16-VSB cable modes. Both curves exhibit a very steep rise
in error rate around their respective 15- and 28-dB SNR
threshold of visible errors (TOV) or threshold of audible
errors (TOA). The TOV and TOA values are identically
defined as 2.5 MPEG packet errors per second, which is
the experimentally determined point where the prototype
receiver would just begin to show visible decoding errors
due to transmission errors. As the SNR value at the input
to the ATSC receiver decreases from high to low, the DTV
receiver will change from an error-free, perfect picture (and
sound) to an all-error, frozen picture (and muted sound) in
less than 1 dB. This is the well-known “digital cliff effect”
that all digital systems exhibit.
Finally, ATSC and others provide recommendations for
both transmitter [20], [21], [24] and receiver [25], [26] per51
Fig. 7. E-VSB preprocessor and multiplexer.
formance parameters as guidelines for both manufacturers
and broadcasters. The recommendations are reasonable to
achieve and allow the objectives of good DTV reception over
varied conditions such as urban, suburban, and rural, whether
using either indoor or outdoor antennas, and with minimal
antenna adjustment.
IV. ENHANCED VSB (E-VSB) SYSTEM
E-VSB [9], [27] is fundamentally a method of adding
further error protection coding to part of the 8-VSB signal.
It is required to simultaneously provide a performance
increase for the E-VSB coded portion, while not degrading
the “normal” or “main” portion used by legacy ATSC receivers. Secondarily, but importantly, E-VSB applications
require additions to the ATSC transport and PSIP standards
to support functions such as synchronization of separate but
related source material in the main and enhanced streams.
The basic technical advantage of the E-VSB stream is an
improvement of at least 6 dB in SNR and interference thresholds. This is obtained in exchange for heavier forward error
correction coding, and therefore at the expense of payload
data rate for the enhanced part of the transmission. Naturally,
designating a portion of the transmitted symbols as enhanced
reduces the bandwidth of the main stream, just as in the case
of multicasting, where each program uses only part of the
19.39-Mb/s ATSC stream.
Applications envisioned for E-VSB include streams unrelated to the main stream, related streams, and synchronized
related streams such as fallback audio and/or video.
Unrelated streams are envisioned for use in carrying
secondary channels or data that can be used by portable or
PC-based devices with nonoptimum antennas, for example,
a “subchannel” carrying stock market information, news,
and weather.
Fallback audio is defined as a duplicate of the main audio
that can be switched to in the receiver when the main signal is
momentarily lost. The aim is to make this switch as seamless
and unnoticeable as possible. This is the most demanding application envisioned, involving all primary and secondary aspects of E-VSB: the physical layer, synchronization of time
stamps for the main and fallback, and enhanced PSIP that an52
nounces the availability of fallback to the enhanced-capable
receiver.
A. E-VSB Coding Details
Because the reaction of legacy receivers to deliberate
coding errors cannot be predicted, the E-VSB coding is
arranged to be effectively cascaded with the normal 8-VSB
coding. As a result, legacy receivers detect a correctly coded
signal (although they do not recognize the payload contents).
The coding is done in two stages. First, an RS code is applied
to the enhanced data, before it is encapsulated in normally
formed MPEG packets and processed by 8-VSB coding.
These packets are transmitted with an MPEG null-packet
header, and therefore are ignored by legacy receivers. The
“preprocessing” to this point is shown in Fig. 7. (N/E in
figures refers to normal/enhanced data or processing-state
flags.)
Second, an enhanced convolutional code is applied, effectively preceding the 8-VSB convolutional code. This results
in a signal that is indistinguishable in all physical characteristics from a normal 8-VSB signal. It has eight equally probable symbol levels and the same spectral characteristics and
power as standard 8-VSB. Fig. 8 shows a block diagram of
a complete E-VSB channel encoder. The blocks in the upper
row are the same as a standard 8-VSB encoder. The blocks in
the lower row are those added for E-VSB coding, including
the preprocessing shown in detail in Fig. 7
There are several details to the processing that are required
to maintain compatibility. In particular, since the enhanced
convolutional code changes the sequence of transmitted
symbols from what it would be for 8-VSB, it invalidates
the 8-VSB RS coding. The E-VSB transmitter therefore
recalculates the RS code to satisfy legacy receivers. Also,
since the 8-VSB convolutional coder is 12-phase, operating
as 12 separate interleaved coders, the E-VSB coder is also
12-phase, and operates in synchronism with the 8-VSB
coder. Fig. 9 shows the concatenation of coders. The inputs
marked N/E are normal/enhanced flags that control the
coding on a symbol-by-symbol basis. When the flags at set
to “N,” normal or “main” data is bypassed around the enhanced coding. When the flags are set to “E,” the equivalent
enhanced coder is as shown in Fig. 10.
PROCEEDINGS OF THE IEEE, VOL. 94, NO. 1, JANUARY 2006
Fig. 8.
Complete E-VSB channel encoder.
Fig. 9.
Concatenated enhanced and standard encoders.
Note that the enhanced portion of the coder contains a
“precoder bypass,” which operates upon the enhanced data.
This effectively removes the interference filter precoding of
the standard coder for enhanced data. This precoder is in the
standard encoder to allow use of an optional comb filter in
the legacy 8-VSB receivers to reduce analog cochannel interference. However, it doubles the number of trellis states
that must be decoded when it is used. The precoder bypass
prevents this expansion in the enhanced trellis decoder. However, because the precoder bypass circuit has two possible
states depending on the history of normal data, the enhanced
BRETL et al.: ATSC RF, MODULATION, AND TRANSMISSION
trellis decoder in new enhanced receivers must determine the
precoder bypass state for each received enhanced symbol.
An added requirement to maintain compatibility with
legacy 8-VSB receivers occurs due to the time multiplexing
of the main and enhanced data. The multiplexing introduces
shifts in the time of emission of the main data packets as
compared to a pure 8-VSB system. This requires compensation of the MPEG program clock references (PCRs) in the
emitted signal, and also some care to maintain adherence to
the MPEG buffer models in the main channel elementary
streams.
53
Fig. 10.
Equivalent encoder for enhanced data.
B. E-VSB Features
E-VSB provides two coding rates designated as “1/2-rate”
and “1/4-rate,” referring to a choice of two convolutional
codes. (The payload is additionally reduced by the ratio
164/188 due to the added RS coding.) This additional
coding provides an SNR threshold advantage of 6 or 9 dB,
respectively, as compared to normal 8-VSB. The arrangement supports applications similar to those of hierarchical
transmission, but because the different types of data are
actually time multiplexed, they do not interact, and the
performance of legacy receivers on the main data is not
affected by the presence of enhanced data.
C. E-VSB Signaling
Because the enhanced data is identical to 8-VSB data in
all physical respects, a means is required for the enhanced
receiver to identify it readily. It would be conceivable to
identify enhanced data by attempting to decode all the data
and accepting successfully decoded data only. This would
result in an unacceptably long decode/decision process and
would preclude using the data to enhance receiver performance in other areas such as the channel equalizer. Therefore, the amount and placement of the enhanced data in a
VSB data field is signaled by “map” data transmitted in the
field sync segment (Fig. 11).
Twelve bits are carried as a 64-bit Kerdock code, which
alternates polarity from field to field, both as protection from
fixed ghost patterns and so that it can be distinguished easily
from a legacy 8-VSB broadcast, which will contain a fixed
data pattern. Nine map bits indicate the proportion of main,
1/2-rate, and 1/4-rate data. Two bits in each data field indicate a 16-field countdown until the next map change. One bit
indicates a choice of two patterns for packing the enhanced
data in a data field.
A particular map number indicates the number of data segments in a data field devoted to 1/2-rate data, the number devoted to 1/4-rate data, and the enhanced segment positions
in the field. However, the enhanced data of both rates passes
through a single “enhanced” byte interleaver and is cut into
164-B chunks, then has the enhanced RS code appended,
then is placed among the main packets, and finally undergoes
the 8-VSB interleaving. Therefore, there is no place in the
54
Fig. 11.
Data field with E-VSB map data.
system where the data is neatly confined to a packet interval
except at the transmitter transport stream input and receiver
transport stream output. The main and enhanced symbols are
thoroughly interleaved on the channel, and the detailed pattern of interleaving must be reconstructed in the receiver in
order to recover the enhanced data.
V. MULTIPLE TRANSMITTER NETWORKS
The conventional approach to covering a large television
service area involves the placement of a single high-power
transmitter at a central location. Under certain conditions,
however, the conventional method may face economical and
technical challenges that require careful considerations and
engineering solutions.
With the single-transmitter configuration, signal levels are
not uniform throughout the service area. The radiated power
of the transmitter is usually calculated so as to provide sufficient signal strength at the edges of the coverage area. In
locations closer to the transmitter, the signal is stronger and
may be considerably more than required for a satisfactory
reception.
The high cost of extending the coverage area of a transmitter by increasing its radiated power is another potential
problem with the single transmitter approach. Serving the
last kilometer of coverage is far more expensive than the first
kilometer. For example, for a UHF digital TV signal at about
PROCEEDINGS OF THE IEEE, VOL. 94, NO. 1, JANUARY 2006
80 km from a transmitter whose antenna is 300 m above average terrain, approximately a 3 dB (or 100%) increase in
transmitted power would be required to increase coverage by
5 km. Thus, increasing coverage with raw transmitter power
can be expensive to accomplish [28]. Another issue is interference into neighboring service areas. Based on calculations
for the location and time availability of F(50, 10) for interference and F(50, 90) for coverage using FCC curves, it can be
shown that cochannel interference from a digital UHF transmitter will extend on the order of three times the distance
over which it can provide coverage. So, extending by 1 km
the coverage area of a single transmitter by increasing its
output power would add 3 km more to its cochannel interference zone.
In situations such as those detailed above, one possible solution is to construct a multiple transmitter network and distribute the signal across the coverage area by using a number
of lower power transmitters instead of a single central transmitter. Among the potential benefits of this approach are
[28]:
• more uniform and higher average signal levels
throughout the service area;
• more reliable outdoor and indoor reception as a result
of higher average signal levels;
• less overall effective radiated power (ERP) and/or antenna height requirements, resulting in less interference;
• stronger signals at the edges of the service area without
increasing interference to neighboring stations.
A multiple transmitter approach of sorts is used for analog
TV systems in the form of translators. Such systems are
mostly used to fill coverage gaps, or to extend the coverage area. They are not usually intended to work with the
main transmitter to uniformly distribute the signal across
the service area. Instead, there is typically a master/slave
relationship between the higher power central transmitter
and the lower power translators. A primary limitation is the
number of RF channels that must be used for an analog network. Usually, for transmitters, channels are required.
A. Distributed Transmission Networks (DTxN)
Distributed transmission can be regarded as a way of
covering a service area with a network of two or more
transmitters, all synchronized and emitting exactly the same
program, and operating according to technical guidelines
and standards specifically developed for this type of system
[9], [28]. The number of channels used can be far less than
the number of the transmitters that constitute the network.
Application of DTxN is not limited to filling coverage gaps
or extending the coverage area, which is usually the case in
analog TV translators. It can also be used for creating a more
uniform distribution of the transmitted signal in the main
parts of the service area, as well as enhancing the signal in
other parts by illuminating the area from different directions.
1) Single-Frequency DTx Networks: If all the transmitters
constituting a DTx network are synchronized and use the
same channel for transmission, they form a single-frequency
network (SFN). SFN is one of the interesting possibilities
BRETL et al.: ATSC RF, MODULATION, AND TRANSMISSION
provided by DTV transmission systems. It is not feasible
with analog TV due to the creation of harmful ghosts.
SFNs for single-carrier signals such as 8-VSB have become possible because of the use of adaptive equalizers in
receivers. When signals from a number of synchronized SFN
transmitters arrive at a receiver, the adaptive equalizer can
treat those signals as echoes of one another and extract the
data they carry. This assumes, of course, that the relative amplitude and delay of the signals fall within the capabilities
of the adaptive equalizer. This is a serious implementation
issue because it must take into account the capabilities of already-deployed receivers.
2) Multiple-Frequency DTx Networks: A multiple-frequency DTx network uses more than one channel for its
synchronized transmitters. Such a network may be composed of a group of smaller embedded SFNs, or a hybrid
of SFNs and individual transmitters operating on channels
other than those of the SFNs.
B. Limitations of DTx Networks
As a tradeoff for their benefits, DTx networks may have to
deal with certain operational restrictions under specific conditions. Such limiting conditions could exist if a single-frequency DTx network and a single transmitter operating on
the adjacent channel coexist in the same market area. Under
these conditions, implementation of the SFN should be based
on a very careful and subtle design to minimize interference
to the single transmitter.
Within-market DTV adjacent channel operation was allowed in DTV planning under the condition of colocating the
corresponding transmitters, or at most separating them by a
distance not more than a specific value (5 km in the United
States). In an SFN environment, only one of the network
transmitters can be colocated with the adjacent channel
transmitter, and for other SFN transmitters, limiting the
output power—or taking other appropriate measures—may
be necessary to avoid unacceptable interference to the adjacent channel transmitter [28].
In the presence of an NTSC adjacent channel, the limitation on operation of an SFN is more serious and may make
the implementation of the network very challenging in the
transition period from NTSC to DTV because of the much
higher protection ratios demanded by NTSC (as compared
to DTV).
ATSC Recommended Practice A/111 [28] provides many
guidelines for designing a DTx network and managing interference under various conditions, including the presence of
DTV or NTSC adjacent channel signals.
The limitations of multiple-frequency DTx networks arise
mainly from any embedded SFNs, which can be designed
according to the same principles as any SFN.
C. Configurations of DTx Networks
ATSC A/111 [28] proposes three methods or structures (or
any of their combinations) for implementing a DTx network.
These structures, however, have their pros and cons, and any
of them can be considered more suitable depending on the
55
specific situation and the conditions under which they are
used.
1) Distributed-Transmitter Network: In this method, a
central studio sends a baseband signal or video/audio data
stream to the DTxN transmitters via studio-transmitter-links
(STL), which can be fiber optic, microwave, satellite, etc.
Frequency and time synchronization of different transmitters constituting the network is based on ATSC A/110, “Synchronization Standard for Distributed Transmission” [29]. A
distributed transmitter network is a single-frequency DTx
network.
2) Distributed-Translator Network: In this method, the
transmitters constituting the network are coherent translators, all operating on the same channel, and translating the
frequency of an over-the-air signal received from a main
DTV transmitter to a second RF channel. This eliminates the
need for an STL and makes frequency and time synchronization of different network transmitters quite simple. In this
method, however, two channels are used, one for the coherent
translator output, and one for the main transmitter feeding
them. One may consider this as a sort of frequency diversity in the overlapping coverage area of the main transmitter
and the translators. A distributed translator network is a multiple-frequency DTx network.
3) Digital On-Channel Repeater (DOCR): Like the
method described in Section V-C2, the transmitters constituting this network pick up their inputs from a main
transmitter, eliminating the need for an STL, but transmit
on the same channel as they receive. With this approach,
two limiting factors exist in the operation of the network.
First, depending on the relative locations of the repeaters
and the main transmitter, long “preechoes” may be created
by the main transmitter in the overlapping coverage areas.
Second, depending on the amount of feedback from the
DOCR transmitting antenna to its receiving antenna, there
will be a power limitation on the repeater output. A network
consisting of DOCRs and a main transmitter is a single-frequency DTx network.
D. Applications of DTx Networks
There are a number of potential applications for DTx networks; some are detailed below, grouped by complexity and
size.
In its simplest form, a single-frequency DTx network can
be formed by simply adding a second transmitter—which
may be an on-channel repeater—to a main transmitter for
filling a gap or extending coverage to include a town that
is beyond the reach of the main transmitter [30]. With this
simple application, no additional channel is used by the network.
In a more complex case, an SFN may be used to improve
service and enhance reception in areas that are already
within the coverage of the main transmitter. For example, in
downtown canyons with shadow areas caused by high-rise
buildings, it may be desirable to construct a network of
distributed transmitters or DOCRs, or a distributed translator
network operating on a second channel, to provide a more
reliable portable or indoor reception environment [31],
56
[32]. Different transmitters in such a supplementary DTx
network can illuminate a common target area from different
directions, giving receivers a better chance of successful
reception.
A DTx network can also be used to replace an operational
or a planned single (central) high-power transmitter with an
SFN comprised of lower power transmitters operating on the
same channel, and possibly some DOCRs. Under these conditions, the operation of the SFN should be within the protected contour, limited by the specifications of the original
transmitter to be replaced by the network. A DTx network
also may be a hybrid of different configurations operating
on more that one channel [33]. Such a hybrid network may
contain many transmitters with different output powers and
covering a very large service area.
E. DTx Design Considerations
The use of DTx networks leads to a range of potential complications that must be addressed in the design process. Key
among the issues to consider is the interference environment,
which comprises two types of interference: external and “network internal” interference.
1) External and Internal Interference in a DTx Network:
For a transmitter in a DTx network, external interference
may be generated by sources such as cochannel NTSC and
cochannel DTV stations that do not belong to the network,
adjacent channel NTSC and DTV stations, and other sources
that are usually involved in creating interference to conventional single (central) transmitters. A single-frequency DTx
environment may also include network internal interference,
which does not exist in the case of the single transmitter configuration.
As there is more than one DTx transmitter emitting the
same signal on the same channel in the service area of a
single-frequency DTx network, a receiver in the overlapping
coverage area may receive signals from different SFN transmitters. The receiver considers the strongest signal as the
main signal and the others as echoes. There should be a distinction between these signals and natural echoes, which are
reflections from stationary and moving objects, and are also
present in a single (central) transmitter environment. In order
to be distinguishable from natural echoes, they may be referred to as “SFN signals.” There is usually no control over
natural echoes but the DTx network designer has control over
the amplitude and time delay of the SFN signals.
In parts of the service area of an SFN that can receive
signal from multiple transmitters, the amplitude and/or delay
spread of the SFN signals may fall beyond the capability
of the receiver equalizer range, and create network internal
interference. Minimizing the creation of such areas, or preventing them from falling over populated areas, is an important design criterion for a single-frequency DTx network.
It is important to note that under many circumstances, creation of multiple SFN signals in specific areas may be a desired and deliberately planned, and can be quite helpful if
the network is operating on the basis of proper design criteria. For example, to enable more reliable portable or indoor reception in a downtown canyon, multiple SFN signals
PROCEEDINGS OF THE IEEE, VOL. 94, NO. 1, JANUARY 2006
(under the right conditions) from different directions may be
a matter of choice [32]. A correctly designed network will
minimize the possibility that the multiple transmitter signals
will result in harmful interference.
2) Managing Interference in DTx Networks: Various
design methods can be applied to managing both external
and network-internal interference in a DTx network. For
example, selecting a specific network configuration that fits
well with the topography of the service area and benefits
from terrain shielding may minimize one or both types of
interference. Antenna directivity may be applied to both
cases, and delay adjustment of the transmitters may also be
applied to mitigate network-internal interference. All such
techniques should be considered when designing a DTx
network [9].
It should be noted, however, that mutual external interference problems in a DTx network can be similar to those of
the single (central) transmitter architecture, and resolved by
the same techniques. As far as network internal interference
is concerned, the most important issues to consider are [9]:
• proper selections of the radiated powers;
• separation distance between the SFN transmitters;
• time adjustment by applying appropriate relative delays.
Incorrect choices for these parameters can make the relative
amplitude and delay of the SFN signals in some parts of the
overlapping coverage areas fall beyond the receiver’s adaptive equalizer capabilities, and can cause reception failure in
those areas.
3) Receiver Constraints: Another important issue affecting the design of a DTx network is receiver performance
with respect to echo handling capabilities. Better receivers,
capable of handling stronger pre- and postechoes, over a
wider range of delays, make DTx network design more
flexible and simpler. On the other hand, receivers with
weaker capabilities put more restrictions on the design and
implementation of DTx networks. As the technology of
adaptive equalizers improves over time, this consideration
will become less stringent, allowing more flexible SFN
designs while maintaining a particular level of reliability.
F. Implementation of Distributed Transmission Networks
A generalized DTx network may include different
combinations of distributed transmitters, distributed translators, and DOCRs. Each of these structures can be implemented by using a number of methods that differ in
their degree of complexity, mostly related to the achieving
synchronization between different network transmitters.
For example, the synchronized translators used in a DTx
network can operate on the basis of RF-RF, or RF-IF-RF
conversion, or the signal can be brought down to base-band
and decoded, error corrected, and then reencoded and upconverted to RF. Selection of each of these approaches
is determined by the existing conditions. For example, if
the input signals received by the translators are strong and
stable, RF-RF operation may be adequate. Under these conditions, two translators could be frequency synchronized by
phase-locking their local oscillators using global positioning
BRETL et al.: ATSC RF, MODULATION, AND TRANSMISSION
system (GPS) signals. Time adjustment can be achieved
by applying appropriate delays to the signals when passed
through the translators [32]. This RF-RF approach is the
simplest and most cost-effective method of implementing
parts of a DTx network consisting of distributed translators
or DOCRs. Synchronization of RF-IF-RF translators or
DOCRs, however, should be based on more sophisticated
methods [33].
In other circumstances, such as when translator signals
are brought down to base-band or when a number of STLs
are used to bring the base-band signal to the transmitters,
achieving frequency and time synchronization becomes
more complex. Under these conditions, the synchronization
process should be based on the methods specified in A/110
[29].
1) Transmitter Identification: As of August 2005, there
were more than 1500 DTV transmitters in operation in
the United States and Canada. As the number of DTV
transmitters grows, there is an increasing need to identify the origin of each DTV signal received at different
locations. ATSC A/110 [29] includes specifications for a
spread spectrum sequence, embedded as a RF watermark,
for transmitter identification (TxID) purposes. Transmitter
identification techniques (or transmitter fingerprinting) are
used to detect, diagnose, and classify the operating status
of radio transmitters. Transmitter identification also enables
broadcast authorities and operators to identify the source(s)
of interference, if any. More importantly, TxID can be used
to tune various transmitters in an SFN to minimize the
effects of multipath interference caused by the destructive
interference of several different transmissions and/or by the
reflection of transmissions.
To allow identification of individual transmitters in a network, provision is made to assign specific, identifiable codes
to particular transmitters [29]. The transmitter identification
codes are combined and used to generate a symbol sequence
that is modulated synchronously with the host 8-VSB symbols in such a way that ordinary receivers cannot detect their
presence but special monitoring and measuring instruments
can [34], [35]. The ATSC RF watermark additionally can be
used for robust data transmission and positioning applications [36]–[39].
REFERENCES
[1] Federal Communications Commission, “Advanced television
systems and their impact on the existing television broadcast
service—Review of technical and operational requirements: Part
73-E, Television broadcast stations—Reevaluation of the UHF
television channel and distance separation requirements of part 73
of the Commission’s rules,” Washington, DC, FCC 87-268: MM
Docket no. 87-268 , adopted 16 Jul. 1987, released 20 Aug. 1987.
[2] F. Box, “A heuristic technique for assigning radio frequencies to
mobile radio nets,” IEEE Trans. Veh. Technol., vol. VT-27, no. 2,
pp. 27–64, May 1978.
[3] D. S. Johnson, C. R. Aragon, L. A. McGeoch, and C. Schevon,
“Optimization by simulated annealing: An experimental evaluation—Part I, Graph partitioning,” Oper. Res., vol. 37, no. 6, pp.
865–892, Nov.–Dec. 1989.
[4] G. A. Hufford, A. G. Longley, and W. A. Kissick, “A Guide to the
use of the ITS irregular terrain model in area prediction mode,” U.S.
Department of Commerce, Washington, DC, NTIA Rep. 82-100,
1982.
57
[5] Terrain Irregular Rough Earth Model (TIREM), NTIA, U.S. Department of Commerce, Washington, DC, “Microcomputer spectrum analysis models (MSAM) Version 4,” 1992.
[6] Federal Communications Commission, Washington, , DC, “Sixth
report and order in the matter of advanced television systems and
their impact on the existing television broadcast service,” FCC
97-115: MM Docket no. 87-268, adopted 3 April 1997, released
21 April 1997.
[7] Federal Communications Commission, Washington, , DC, “Memorandum opinion and order on reconsideration of the sixth report
and order in the matter of advanced television systems and their impact on the existing television broadcast service,” FCC 98-24: MM
Docket no. 87-268, adopted 17 February 1998, released 23 February 1998.
[8] Federal Communications Commission, Washington, , DC, “FCC
OET Bulletin 69,” Jun. 2, 1997.
[9] ATSC digital television standard, ATSC A/53D, Advanced Television Systems Committee, Washington, DC, Jul. 19, 2005.
[10] Guide to the use of the digital television standard, ATSC A/54A,
Advanced Television Systems Committee, Washington, DC, Dec.
4, 2003.
[11] Federal Communications Commission, Washington, DC, “Fourth
report and order,” FCC 96-493, MM Docket no. 87-268, Dec. 24,
1996.
[12] Federal Communications Commission, Washington, DC, “Fifth report and order,” FCC 97-116, MM Docket no. 87-268, Apr. 3, 1997.
[13] Federal Communications Commission, Washington, DC, “Sixth report and order,” FCC 97-115, MM Docket no. 87-268, Apr. 3, 1997.
[14] Code of Federal Regulations, Telecommunications, , Title 47, Ch.
I, Part 73, Subpart E, Sec 73.622–73.625; Part 76, Subpart K,
Technical Standards, Federal Communications Commission, Oct.
1, 1997, (revised October 1, 2003).
[15] G. Sgrignoli, “Measuring peak/average power ratio of the
Zenith/AT&T DSC-HDTV signal with vector signal analyzer,”
IEEE Trans. Broadcast, vol. 39, no. 2, pp. 255–264, Jun. 1993.
[16] International Telecommunication Union (ITU-T), “Digital multiprogramme systems for television, sound, and data services for
cable distribution,” Recommendation J.83, Apr. 1997.
[17] International Telecommunication Union (ITU-T), “Distribution of
digital multi-programme signals for television, sound, and data
services through SMATV networks,” Recommendation J.84, Mar.
2001.
[18] Federal Communications Commission, Washington, DC, “Memorandum opinion and order on reconsideration of the sixth report and
order,” FCC 98-24, MM Docket no. 87-268, Feb. 17, 1998.
[19] G. Sgrignoli, “DTV repeater emission mask analysis,” IEEE Trans.
Broadcast, vol. 49, no. 1, pp. 32–80, Mar. 2003.
[20] Advanced Television Systems Committee, Washington, DC,
“Transmission measurement and compliance standard for digital
television,” ATSC A/64A, May 30, 2000.
[21] C. Eilers and G. Sgrignoli, “Digital television transmission parameters—Analysis and discussion,” IEEE Trans. Broadcast, vol. 45,
no. 4, pp. 365–385, Dec. 1999.
[22] G. Sgrignoli, W. Bretl, and R. Citta, “VSB modulation used for
terrestrial and cable broadcasts,” IEEE Trans. Consum. Electron.,
vol. 41, no. 3, pp. 367–382, Aug. 1995.
[23] W. Bretl and G. Sgrignoli, “Summary of the Grand Alliance VSB transmission system laboratory tests,” Jun. 1996
[Online]. Available: http://www.zenith.com/digitalbroadcast/
downloads/ATSC%20Lab%20Test%20Results.pdf
[24] G. Sgrignoli, “Interference analysis of co-sited DTV and NTSC
translators,” in IEEE Broadcast Symp. Handouts, Washington,
D.C., 2004.
[25] Receiver performance guidelines, ATSC A/74, Advanced Television Systems Committee, Washington, DC, Jun. 18, 2004.
[26] G. Sgrignoli, “Preliminary DTV field test results and their effects
on VSB receiver design,” IEEE Trans. Consum. Electron., vol. 45,
no. 3, pp. 894–915, Aug. 1999.
[27] E-VSB implementation guideline (work in progress), ATSC
T3S9-2-014, Advanced Television Systems Committee, Washington, DC.
[28] Design of synchronized multiple transmitter networks, ATSC
A/111, Advanced Television Systems Committee, Washington,
DC, Sep. 3, 2004.
[29] Synchronization standard for distributed transmission, ATSC
A/110A, Advanced Television Systems Committee, Washington,
DC, Jul. 19, 2005.
58
[30] K. Salehian, M. B. Guillet, B. Caron, and A. Kennedy, “Onchannel repeater for digital television broadcasting service,” IEEE
Trans. Broadcast, vol. 48, no. 2, pp. 97–102, Jun. 2002.
[31] K. Salehian, B. Caron, and M. Guillet, “Using on-channel repeater to improve reception in DTV broadcasting service area,”
IEEE Trans. Broadcast, vol. 49, no. 3, pp. 309–313, Sep. 2003.
[32] K. Salehian, Y. Wu, and B. Caron, “Design procedures and field
test results of a distributed-translator network, and a case study for
an application of distributed-transmission,” presented at the National Association of Broadcasters Conf. (NAB2005), Washington,
DC.
[33] Y. T. Lee, S. I. Park, H. M. Eum, H. N. Kim, S. W. Kim, and S. I.
Lee, “A novel on-channel repeater for single frequency network in
ATSC system,” in Proc. Nat. Association Broadcasters Conf. 2004,
pp. 128–133.
[34] A. Mattsson, M. Weiss, M. Simon, D. Hershberger, Y. Wu, and
X. Wang, “Transmitter identification techniques for distributed
transmission networks,” presented at the IEEE Broadcast Technology Soc. 53rd Annu. Symp., Washington, DC, 2003.
[35] X. Wang, Y. Wu, and B. Caron, “Transmitter identification using
embedded pseudo random sequences,” IEEE Trans. Broadcast, vol.
50, no. 3, pp. 244–252, Sep. 2004.
[36] X. Wang, Y. Wu, and B. Caron, “Transmitter identification in
distributed transmission network and its potential application,”
presented at the Nat. Association Broadcasters Conf. (NAB2005),
Washington, DC.
[37] X. Wang, Y. Wu, B. Caron, and J.-Y. Chouinard, “A new position
location system using ATSC TxID signals,” presented at the IEEE
Vehicular Technology Conf., Stockholm, Sweden, 2005.
[38] X. Wang, Y. Wu, and J. Y. Chouinard, “Robust data transmission using the transmitter identification sequences in ATSC DTV
signals,” IEEE Trans. Consum. Electron., vol. 51, no. 1, pp. 41–47,
Feb. 2005.
[39] M. Rabinowitz and J. J. Spilker, Jr., “A new positioning system
using television synchronization signals,” IEEE Trans. Broadcast,
vol. 51, no. 1, pp. 51–61, Mar. 2005.
Wayne Bretl (Senior Member, IEEE) received
the BSEE from Illinois Institute of Technology in
1966.
He joined Zenith Electronics in 1975. He is
a Principal Engineer in the R&D Department,
Zenith Electronics, Lincolnshire, IL. He holds
over 15 patents in television technology and
related areas.
Mr. Bretl is a member of the Society of Motion Picture and Television Engineers, the Audio
Engineering Society, and the Society for Information Display, and represents Zenith in ATSC and a number of professional
and industry associations.
William R. Meintel received the B.S.E.E. degree
from the West Virginia Institute of Technology,
Montgomery, in 1969.
He has 35 years’ experience in the communications field. He was a Field Engineer for the
Federal Communications Commission (FCC),
Buffalo, NY, and later was Senior Field Engineer
in the Chicago, IL, field office before joining the
FCC’s Media Bureau Policy and Rules Division.
During his tenure in the Media Bureau, he was
extensively involved in a number of complex
domestic and international spectrum planning matters and was a member of
the U.S. delegation to a number of international conferences. Since entering
private practice, he has been heavily involved in technical consulting
and spectrum planning for the broadcast industry. During that period he
coauthored a report for the NAB on spectrum requirements for digital audio
broadcasting (DAB), created a plan for independent television broadcasting
for Romania, and has been extensively involved in spectrum planning for
digital television (DTV) in both the United States and internationally. His
involvement in DTV planning has included a wide variety of activities
including computer modeling, engineering consulting, and field measurements programs used to validate the proposed systems. More recently he
has been providing considerable engineering support for various clients
PROCEEDINGS OF THE IEEE, VOL. 94, NO. 1, JANUARY 2006
in a number of system design and implementation projects. He has also
authored a number of papers and articles and made numerous presentations
on subjects related to spectrum planning.
Gary Sgrignoli received the B.S. and M.S. degrees in electrical engineering from the University of Illinois, Champaign-Urbana, in 1975 and
1977, respectively.
He joined Zenith Electronics Corporation in
January 1977, where he worked as an engineer
in the Research and Development department
for 27 years. In March 2004, he set up Sgrignoli
Consulting, a DTV-transmission consulting
firm, and in April 2005 he merged his practice
with those of W. Meintel (Techware, Inc.) and
D. Wallace (Wallace and Associates) to create Meintel, Sgrignoli, and
Wallace (MSW), Waldorf, MD. Further information can be found at
http://www.MSWdtv.com. He has worked in the R&D design area on
television “ghost” canceling, cable TV scrambling, and cable TV two-way
data systems before turning to digital television transmission systems. Since
1991, Gary has been extensively involved in the VSB transmission system
design, its prototype implementation, the ATTC lab tests in Alexandria, VA,
and both ACATS field tests in Charlotte, NC. He was also involved with
the DTV Station Project in Washington, DC, helping to develop DTV RF
test plans. He has also been involved with numerous television broadcast
stations around the country, training them for DTV field testing and data
analysis, and participated in numerous DTV over-the-air demonstrations
with the Grand Alliance and the ATSC, both in the United States and
abroad. In addition to publishing technical papers and giving presentations
at various conferences, he has held numerous digital VSB transmission
system seminars around the country. He holds 35 U.S. patents.
Xianbin Wang (Member, IEEE) received the
Ph.D. degree in electrical and computer engineering from National University of Singapore,
Singapore, in 2000.
He was with Institute for Infocomm Research,
Singapore (formerly known as Centre for Wireless Communications), as a Senior R&D Engineer in 2000. From December 2000 to July 2002,
he was a System Designer at STMicroelectronics,
Inc., Ottawa, ON, Canada. Since July 2002, he
has been with Communications Research Centre
Canada, Ottawa, where he is currently a Senior Research Scientist. He is
also an Adjunct Associate Professor of Laval University, QC, Canada. His
current research interests include digital signal processing, broad-band wireless system, and communication theory.
BRETL et al.: ATSC RF, MODULATION, AND TRANSMISSION
S. Merrill Weiss is a graduate of the Wharton
School, University of Pennsylvania, Philadelphia.
He is a consultant in electronic media
technology, technology management, and management. He conducted the experiments that
led to the very first digital television standard
(CCIR Recommendation 601) in 1981. He was
a major contributor to the work of the FCC
Advisory Committee on Advanced Television
Service, doing the bulk of the work on implementation. He currently participates in the Advanced Television Systems
Committee’s (ATSC’s) efforts. In the original ATSC standard for 8-VSB,
there was no method available for synchronizing transmitters as needed
for single-frequency networks; he invented a synchronization method that
is now embodied in the recently adopted ATSC standard on transmitter
synchronization. He has one issued patent and one pending.
Mr. Weiss has been recognized by the Society of Motion Picture and Television Engineers by elevation to the rank of Fellow and was the 1995 recipient of its David Sarnoff Gold Medal and the 2005 recipient of its Progress
Medal. He was also nominated for a technical Emmy Award for his conception of serial digital interfaces for television systems. He has been certified
by the Society of Broadcast Engineers at the level of Professional Broadcast
Engineer. He is a member of the IEEE Broadcast Technology Society.
Khalil Salehian received the B.Sc. and M.Sc.
degrees in electrical engineering from Pahlavi
University, Tehran, Iran, in 1971 and 1974,
respectively, and completed graduate studies
in computer science at Concordia University,
Montreal, QC, Canada, in 1999.
He spent about 20 years working with National
Iranian Radio and Television (NIRT) in the fields
of radio and TV network expansion, frequency
planning, and coordintaion and compatibility
with other services. Since 2000, he has been with
the Communications Research Centre Canada (CRC), Ottawa, ON. He is
currently working with the Television Systems and Transmission group
of CRC. His previous fields of interest included radio and TV network
expansion, frequency planning, and coordination and compatibility with
other services. His current fields of interest include computer simulations
and coverage predictions of DTV systems, distributed transmission, and
single-frequency networks for DTV broadcasting services. His work on
distributed transmission systems earned him the 2003 IEEE Scott Helt
memorial award.
59
Download