`
`Modulation Schemes: Moving Digital Data With Analog Signals | EE Times
`
`
`
`
`
`DESIGNLINES | AUTOMOTIVE DESIGNLINE
`Modulation Schemes: Moving Digital Data With Analog Signals
` 0
`By Andrew W. Davis, 10.04.97
`
`
`Modulations are the techniques to carry digital data over analog waveforms. This rather arcane subject
`has been brought to the forefront of the DSP, EE, and telecommunications worlds by the ongoing
`interest in broadband communications, specifically, the great opportunity represented by bringing low
`cost broadband communications to the home. Think what it would be like to log on to the Internet or to
`the corporate LAN at speeds of over a megabit per second. The potential is enormous. The phone
`companies are approaching this opportunity with a grab bag of technologies known as DSL, digital
`subscriber loops. Many articles have appeared recently on ADSL, RDSL, SDSL, HDSL, MDSL, and
`VDSL. These will be covered in more depth in future newsletters. The cable companies are offering
`broadband services with cable modems. In either case, modulations are one of the fundamental
`technologies.
`
`The modulation process places (analog or digital) signal information onto sinewave carriers while
`demodulation reverses the process at the receiving end. Modulation schemes are very much in the
`news today because newer algorithms that take advantage of newer and more powerful DSP
`architectures make possible faster and more reliable communications than was possible before.
`However, modulation changes, with few exceptions, are incompatible with previous schemes, making
`the economic cost of improvements very high if there is an installed base of users or equipment to worry
`about.
`
`As you will see below, there are many types of modulation schemes available today. In the xDSL
`marketplace, there is an active marketing war going on between those in the CAP camp and those
`vendors in the DMT camp. A companion article in this newsletter from Rupert Baines of Analog Devices
`goes a long way towards explaining the CAP vs. DMT debates. Another technology camp where
`modulations are very much in the news is the cable modem camp. The cable operator companies have
`banded together to sort out these issues in order to bring a measure of standardization to their industry.
`These standardization issues are outside the scope of this newsletter. The information below is an
`overview to help you sort through the issues.
`
`Analog modulation processes perform their magic by changing one or more of the three characteristics
`of a sine wave: amplitude, frequency, and phase.
`
`Digital modulation applies a digital data stream to the carrier and makes the data stream compatible with
`the RF communications channel. Each wave state generated in this way represents one symbol of data
`(each symbol is an N-bit word where N is a power of two from 1 to 8, depending on the technology
`used). The resulting modulations schemes are called amplitude shift keying, frequency shift keying, and
`phase shift keying. In RF communications however, the two main approaches are phase shift (constant
`amplitude) and amplitude shift. The number of symbols per second transmitted is known as the baud
`rate. The number of bits per second equals (symbols per second) multiplied by (bits per symbol).
`
`https://www.eetimes.com/document.asp?doc_id=1255090
`
`1/11
`
`00001
`
`Ex. 2002
`T-Mobile USA, Inc. v. Intellectual Ventures II LLC
`IPR2018-01775
`
`
`
`Modulation Schemes: Moving Digital Data With Analog Signals | EE Times
`12/19/2018
`When it comes to modulations, the lack of standards becomes quite evident. A hodgepodge of
`modulation techniques with a range of price/performance features are in use today, although the cable
`modem industry seems to be settling on a de-facto turn to a 64-QAM (N=6) or 256-QAM (N=8) delivery
`model for downstream data, and QPSK for a moderate bit-rate return path.
`
`All modulation schemes can be judged by their spectral efficiency and by their error rates. Spectral
`efficiency is the input digital rate divided by the allocated RF channel bandwidth. The unit of measure is
`"bits/Hz." The error rate (usually failed bits per million or bits per billion of "good" bits) is a function of
`several factors, including susceptibility to noise and interference, susceptibility to fading, and non-
`linearities, which can arise due to dependencies on signal frequency and amplitude. In general, as
`spectral efficiency increases, so unfortunately does the error rate, which means a higher signal-to-noise
`ratio might be needed to achieve acceptable error rates.
`
`There are several ways to achieve more than one bit/Hz of throughput. Instead of simple binary
`encoding, the system can define four different voltages or phases for a single wave cycle, allowing one
`cycle to represent a two-bit symbol. If both phase and amplitude can vary simultaneously over four
`values, then one cycle can represent one of 16 discrete logical states. This squeezes 4 bits of data into
`a single wave cycle, or 4 bits/Hz. Much of the work on modulation techniques currently benefiting the
`cable modem industry stems from interest in digital video and from technology developed for telephony.
`While MPEG-2 is becoming the digital video standard for the broadcast industry, digital modulation
`techniques allow vendors to squeeze 6 or more digital channels into the 6 MHz space normally used for
`a single analog channel. This is one of digital video's major benefits (and is the basis for much of the talk
`about 500 cable channels in the future).
`
`The set of available transmission symbols in a particular modulation scheme is known as its alphabet
`while a graph of the alphabet on a complex plane is known as the constellation (see examples below).
`After symbols have been formed and converted to complex numbers, the constellation diagram is drawn
`by plotting the real part, I, and complex (or imaginary) part, Q, on a 2-D map.
`
`
`Carrierless Amplitude Modulation/Phase Modulation (CAP)
`CAP is a bandwidth-efficient two-dimensional passband line code derived from QAM by AT&T Bell Labs
`as part of an effort to produce a variant of QAM that could be efficiently implemented on a digital signal
`processor. 16-CAP has been adopted by DAVIC, the Digital Audio-Visual Interoperability Council, for
`interactive TV and video-on-demand applications and is proposed for SVD systems. In 16-CAP, blocks
`of four bits are mapped into one of 16 possible 2-D symbols in each symbol period. Two bits represent
`the quadrant, two bits identify a symbol within the quadrant. Increasing the number of bits per symbol
`increases bandwidth efficiency. But the modulation scheme also becomes more sensitive to noise. This
`is the basis of any modulation tradeoff. In Switched Digital Video systems, 16-CAP squeezes 51.84
`Mbps into a downstream signal occupying approximately 20 MHz of bandwidth.
`
`
`Code Division Multiple Access (CDMA)
`CDMA is a form of spread spectrum transmission which works by coding and spreading the information
`to be transmitted over a wide band. CDMA is asynchronous, and typically uses a 30 MHz bandwidth.
`CDMA used by some vendors, like Zenith and Cisco, who believe the spread spectrum approach to be
`superior in noisy upstream environments. Maximum digital bandwidth is approximately 10 Mbps over
`cable.
`
`
`Coded Orthogonal Frequency Division Multiplexing (COFDM)
`COFDM is an experimental approach, intended for broadcast TV, which works by taking the transmitted
`data and spreading it over a large number of carriers, rather than modulating it all onto a single carrier.
`
`https://www.eetimes.com/document.asp?doc_id=1255090
`
`2/11
`
`00002
`
`
`
`Modulation Schemes: Moving Digital Data With Analog Signals | EE Times
`12/19/2018
`Hence, COFDM creates a large number of parallel paths, each of which carries data at a slower rate
`than the overall signal. The longer symbol times are more resistant to system noise. The data on the
`carriers can be modulated using one of the standard digital modulation schemes such as QPSK, 16-
`QAM, 64-QAM, or 256-QAM. Data is spread redundantly over many carriers so that a loss of some
`carriers leads only to loss of an occasional bit, a problem which can be corrected for at the receiving
`(forward-error correction) end. ADC Telecom uses orthogonal frequency division multiplexing for its
`cable-based telephony product to modulate 240 individual DS0 channels into a 5MHz spectrum with 0.5
`MHz guardband on either side. This is very attractive to a cable operator with only 18 MHz usable
`upstream bandwidth. While OFDM has many advantages, it requires more signal processing
`horsepower at the headend than do some of the other modulation techniques.
`
`Figure 1: OFDM breaks the channel into many subchannels and shifts traffic to a clear channel when
`needed.
`
`
`Quadrature Amplitude Modulation (QAM)
`QAM systems combine PSK and ASK to increase the number of states per symbol. QAM is a proven
`technique for the transmission of digital data over a wide range of channels from voiceband modems at
`9600 bps to microwave links transmitting hundreds of Mbps. QAM is also the modulation technique used
`by V.34 modems. Each symbol value represents multiple bits. 16-QAM carries 4 bits per symbol while
`256-QAM carries 8 bits per symbol. The signal-to-noise ratio at the receiver determines the QAM level
`that can be used reliably on a given transmission channel. Typical terrestrial and cable channels allow
`16-QAM and 256-QAM, leading to digital data rates of approximately 20 and 40 Mbps, respectively.
`
`In a typical cable TV application, 64-QAM can squeeze a 30 Mbps data stream into a 6 MHz (bandwidth)
`TV channel. QAM is also of use for digital video broadcast. Note that for video-on-demand or MPEG-
`based broadcast video delivery, 64-QAM allows five channels of 6 Mbps video for each analog channel
`allocation. For telephony, which uses 64 kbps data streams, a single "video channel" could handle over
`450 downstream phone calls, which would be time-division multiplexed within the datastream. Hence, in
`750 MHz cable systems, the upper 240 MHz can contain up to 1000 3-Mbps datastreams, each carrying
`a unique digital address that directs it to a particular set-top box or cable modem (used for video on
`demand, or VOD). QAM is used in some upstream traffic designs, but is less noise resistant, though
`more bit-efficient, than QPSK.
`
`It is easier to visualize QAM by looking at 16-QAM. QAM separates points widely and is hence fairly
`noise immune. The system for 16-QAM combines 4 input bits to produce 1 signal burst. Both phase and
`amplitude are modulated. Odd-numbered bits in the input stream are combined in pairs to form one of 4
`levels which modulate the sine term. Even-numbered bits are similarly combined to modify the cosine
`term. Sine and cosine terms are then combined.
`
`https://www.eetimes.com/document.asp?doc_id=1255090
`
`3/11
`
`00003
`
`
`
`12/19/2018
`
`Modulation Schemes: Moving Digital Data With Analog Signals | EE Times
`V(t) = x(t) cos1/2t + y(t)sin1/2t
`
`
`
`
`Figure 2: I-Q diagram or constellation for 16-QAM scheme.
`
`16-QAM has better spectral efficiency than 8-PSK and is less sensitive to noise than 16-PSK because
`the spacings between symbols are larger (see diagrams below). This is true because the symbols are
`not all on the same circle; the resultant signals are not all of the same amplitude.
`
`Figure 3: Digital multiplexing of data, video, and voice services in the cable headend.
`
`
`Quadrature Phase Shift Keying (QPSK)
`QPSK (which is QAM without an amplitude component) has become the preferred modulation format for
`the upstream. QPSK is inherently robust and economical. While other modulation techniques have been
`proposed with efficiencies higher than the 1.5 bits/Hz of QPSK, these formats have yet to be tested as
`thoroughly. QPSK has been selected by DAVIC as the upstream modulation format, and is the current
`front-runner for selection by the IEEE 802.14 committee. QPSK is also used in many satellite systems.
`
`QPSK involves channel hopping until a clear path is found. QPSK is sometimes called four-phase PSK;
`the phase of the carrier can take on one of 4 values. Each transmitted symbol represents two bits.
`
`00
`01
`
`Acos(wt)
`Acos(wt+90)
`
`https://www.eetimes.com/document.asp?doc_id=1255090
`
`4/11
`
`00004
`
`
`
`12/19/2018
`
`10
`11
`
`Modulation Schemes: Moving Digital Data With Analog Signals | EE Times
`Acos(wt+180)
`Acos(wt+270)
`
`
`
`
`Figure 4: Constellations for QPSK (4-PSK) and 8-PSK
`
`The system is less bit-efficient, but more noise-resistant and has advantages in its ability to operate over
`long distances with many interfering sources such as those found in a neighborhood cable network. For
`this reason, QPSK is favored for upstream traffic in cable modem applications. QPSK delivers about 1.5
`bits per Hertz of bandwidth used. QPSK can go up to 10 Mbps in cable systems, but uses up a large
`portion of the available upstream bandwidth. Hence, most symmetrical cable modem products are
`expected to be limited to 10 Mbps. QPSK is also used for cable-based telephony applications (as well as
`some set-top box designs). Approximately 50 kHz of bandwidth is required for one DS0 channel (64
`kbps). Individual channels are assigned to callers on a per-call basis anywhere within the 6-to-42 MHz
`band (downstream telephony modulations are different. Up to 72-64 kbps DS0 channels are packaged
`within a single 3 MHz slot in the 50-to-750 MHz region).
`
`https://www.eetimes.com/document.asp?doc_id=1255090
`
`5/11
`
`00005
`
`
`
`12/19/2018
`
`Modulation Schemes: Moving Digital Data With Analog Signals | EE Times
`
`Figure 5: Block diagram of cable modem with QPSK modulation. Source: Hewlett Packard
`
`
`Synchronous Code Division Multiple Access (S-CDMA)
`S-CDMA is a modulation scheme introduced by Terayon Corporation. With ordinary time division
`multiple access (TDMA) modulations, users share data channels by taking turns accessing the network
`in different time slots; with ordinary frequency division multiple access (FDMA) modulations, all the time
`is used but the available frequencies are divided up into multiple channels. With S-CDMA technology,
`the information is a spread over a wide band of the spectrum and all the frequencies are used all the
`time. S-CDMA allows for 10 Mbps throughput over each 6 MHz channel, upstream and downstream.
`The spread spectrum results in 10 Mbps by sending multiple streams of data, each comprised of 64
`kbps. These are interleaved within a 6 MHz bandwidth. Individual 64 kbps streams may be allotted to
`telephony, while multiples of these may be used for videoconferencing, Internet access, etc. S-CDMA,
`according to Terayon, addresses the cable modem noise ingress problem better than a frequency-
`hopping scheme because it eliminates the process of searching for clean frequencies. S-CDMA spreads
`the information over the entire 6 MHz channel and uses encoding to make the transmission noise
`immune. Also, according to Terayon, the 10 Mbps upstream bandwidth addresses the capacity issues
`presented by the limited 5-42MHz spectrum.
`
`
`Vestigal Side Band (VSB)
`VSB is the major competing modulation technique (to QAM) for downstream transmission on HFC
`networks. 32-VSB is one of the modulations used by Zenith for downstream delivery. 2-, 4-, and 8-VSB
`is also used by Hybrid Networks for upstream data. Using VSB, operators can offer reverse-band
`channels at 512 kbps using a 300 kHz channel, or they can range speeds between 128 kbps and 2.048
`
`https://www.eetimes.com/document.asp?doc_id=1255090
`
`6/11
`
`00006
`
`
`
`Modulation Schemes: Moving Digital Data With Analog Signals | EE Times
`12/19/2018
`Mbps. Data traveling upstream can be rapidly moved to cleaner areas of the 5-40 MHz spectrum during
`adverse conditions posed by electrical noise and signal ingress. VSB is the modulation scheme selected
`by the Grand Alliance for digital television.
`
`
`QAM and VSB Comparison
`The March 1995 issue of the IEEE Transactions on Broadcasting carried a paper written by K. Kerpez of
`Bellcore which compares QAM and VSB for HFC networks. The paper notes that both modulations are
`bandwidth efficient and use multiple signal levels to send multiple bits/Hz. VSB conserves bandwidth by
`only transmitting a single sideband of the modulated RF spectrum while QAM conserves bandwidth by
`sending two orthogonal sine and cosine carriers in the same frequency band. The paper concludes that
`the two approaches have practically the same overall performance and costs on an HFC network,
`although they are incompatible and have many differences.
`
`
`
`16-VSB
`
`64, 256-QAM
`
`Proponents
`
`Zenith, Grand
`Alliance
`
`Broadcom, AT&T, Scientific-Atlanta,
`General Instruments
`
`Prior Use
`
`TV
`
`Receiver
`
`Analog front-end
`demodulator
`
`Modems
`
`All digital
`
`Symbol
`Rate
`
`Information
`Bit Rate
`
`10.76 Mbaud
`
`5 Mbaud
`
`38.6 Mbps
`
`27 Mbps (64-QAM)
`36 Mbps (256-QAM)
`
`Table 1: Comparison of VSSB and QAM. Source: IEEE Transactions on Broadcasting, Vol. 41, No. 1,
`March, 1995, page 9
`
`
`Modulation and Testing
`In a paper presented on Testing of Digital Video on Cable TV Systems, engineers from Hewlett Packard
`noted that a cable system will probably have to manage program material transported in different digital
`video formats. This presents a myriad of electronics testing issues to the headend system operator, a
`subject outside the scope of this study. However, the paper did present the following comparison of
`cable modulation techniques.
`
`Modulation
`
`Advantage
`
`Disadvantage
`
`QAM
`
`VSB
`
`High spectral
`efficiency
`
`Sensitive to signal-to-noise ratio
`
`Robust carrier and High peak-to-average power ratio
`
`https://www.eetimes.com/document.asp?doc_id=1255090
`
`7/11
`
`00007
`
`
`
`12/19/2018
`
`Modulation Schemes: Moving Digital Data With Analog Signals | EE Times
`symbol clock
`recovery
`
`QPSK
`
`COFDM
`
`Robust in low
`signal-to-noise
`environment
`
`Robust in high
`multipath
`environments
`
`Not spectral efficient
`
`Complex to implement and requires
`more expensive modulation
`hardware
`
`Table 2: Comparison of modulation techniques. Source: Hewlett Packard
`
`
`Modulation and Filtering
`While an FCC channel is defined as having a 6 MHz bandwidth, the edges of this are often not used in
`order to provide "space" between channels and to avoid interference. While an ideal pulse can be used
`to transmit signals, an ideal pulse is impractical. Waveforms in reality are never perfect in shape.
`Practical systems use pulses with more bandwidth than the ideal. The bandwidth above the minimum is
`called excessive bandwidth, usually expressed as a per cent, and involves the use of rolloff filters. Using
`the Nyquist principle, 100% excess bandwidth is twice (2x) the minimum needed. Practical systems
`typically have excess numbers between 10% and 100%.
`
`Figure 6: Bandwidth and rolloff.
`
`Increasing the excess bandwidth simplifies implementation of the communications system, but reduces
`spectral efficiency. The type of rolloff filter used is a network design consideration, and affects the
`throughput possible in a data delivery system. This also explains why there is variation in cable modem
`specifications from different vendors using the same modulation techniques.
`
`
`
`Bits/Hz
`
`Bits/Hz
`
`Bits/Hz Mbps Mbps
`
`Rolloff Filter
`
`0%
`
`QPSK
`
`16-QAM
`
`64-QAM
`
`2
`
`4
`
`6
`
`20%
`
`1.67
`
`3.33
`
`5
`
`100%
`
`20%
`
`100%
`
`1
`
`2
`
`3
`
`10
`
`20
`
`30
`
`2
`
`4
`
`6
`
`https://www.eetimes.com/document.asp?doc_id=1255090
`
`8/11
`
`00008
`
`
`
`12/19/2018
`256-QAM
`
`8
`
`Modulation Schemes: Moving Digital Data With Analog Signals | EE Times
`6.67
`4
`40
`8
`
`Table 3: Effect of filtering on modulation techniques and bandwidth.
`
`Figure 7: Block diagram of cable modem with QAM/QPSK modulation
`
`Jumping quickly from the cable side to the telephone side of the coin, ADSL is still in its formative
`stages. The CAP-based ADSL implementations have enabled 1.5 Mbps MPEG-1 video over common
`two-wire subscriber telephone connections with 64kbps upstream. CAP uses a single QAM signal
`instead of dividing the channel into multiple tones, and is thus more susceptible to interference. A newer
`version of the CAP ADSL technology is being marketed under the GlobeSpan label with one-way
`speeds of over 6 Mbps.
`
`The CAP vs. DMT debate has taken on the nature of a religious holy war within the ADSL community
`and has stolen much attention and, perhaps, momentum. The CAP crowd has been pushing to make
`CAP a standard, pointing out that their technology is less complex, cheaper, easier to implement, and
`meets the large majority of user needs. The DMT team argues that their solution is technologically
`superior from speed and noise immunity considerations. The two are totally incompatible. Several "box
`vendors" have announced that they will have products which support both flavors, so they are
`"modulation neutral" and one vendor, 3Com/US Robotics, has even announced the intention to use a
`very powerful DSP inside their DSL modem which would be capable of running both modulations (only
`one at a time, however) depending on the software code which is downloaded to the chip.
`
`According to an opinion expressed by Analog Devices (see companion newsletter article), resolving "the
`bitter dispute over ADSL line code modulation is crucial in order to expedite system development, speed
`deployment, and ultimately abate the growing traffic congestion on the nation's telephone infrastructure."
`
`CAP is closely related to QAM; in fact the two are compatible. QAM is well understood. It is a single
`carrier signal where the data rate is divided into two and modulated onto two orthogonal carriers I and Q
`using sine and cosine mixers, before being combined and transmitted. CAP operates in the frequency
`domain, whereas DMT operates in the time domain.
`
`DMT is a multi-carrier modulation system that resembles OFDM. DMT divides signal frequency into
`many discrete bands or sub-channels. These are independently modulated with a carrier frequency
`corresponding to the center frequency of the bin and then processed in parallel. DMT's ANSI T1.413
`standard specifies 255 sub-carriers, each with a 4 kHz bandwidth. Each channel can be independently
`modulated from zero to a maximum of 15 bits/Hz (32-QAM on each channel). This allows up to 60 kbps
`per tone. (DMT-based ADSL uses 249 channels for downstream data, leading to a theoretical maximum
`of (249 * 60) 14.94 Mbps. At low frequencies, where copper wire attenuation is low and SNR is good, 10
`
`https://www.eetimes.com/document.asp?doc_id=1255090
`
`9/11
`
`00009
`
`
`
`Modulation Schemes: Moving Digital Data With Analog Signals | EE Times
`12/19/2018
`bits/Hz is typical. In unfavorable line conditions modulation can be relaxed to accommodate lower SNR,
`and 4 bits/Hz is common. The whole process is analogous to simultaneously running 256 modems on a
`chip. The result is 6 Mbps performance on a 4 kHz phone line.
`
`Figure 8: CAP and DMT Concepts.
`
`Conceptually, CAP symbols last a short time (0.001 sec) but have sizable bandwidth. DMT symbols last
`a long time (0.250 ms) but occupy a narrow frequency band. The long time period makes them less
`susceptible to wideband noise spikes.
`
`DMT
`
`CAP
`
`Directs information to
`subcarriers which are
`modulated independently
`
`Single-carrier system
`
`More complex to initialize
`
`Faster start-up time
`
`Rate adaptive, steps of 32
`kbps
`
`Supports rate adaption by varying the
`constellation and the bandwidth in steps
`of 320 kbps
`
`High latency
`
`Low latency
`
`More adept at coping with
`multiple RFI sources
`
`More resistant to RFI, which is
`averaged across the wideband of a
`single carrier
`
`Greater immunity to impulse
`noise because its symbols are
`longer
`
`
`
`More difficult for echo
`cancellation
`
`Lower power needs, simpler analog
`design stages needed
`
`More versatile, flexible, but
`more complex
`
`Easier to implement
`
`https://www.eetimes.com/document.asp?doc_id=1255090
`
`10/11
`
`00010
`
`
`
`12/19/2018
`
`
`Modulation Schemes: Moving Digital Data With Analog Signals | EE Times
`More field experience and test
`equipment available for QAM/CAP
`technology
`
`Patented technique with
`serious intellectual property
`rights questions
`
`Patented technique with serious
`intellectual property rights questions
`
`Few chip sets available
`
`1.5 Mbps chip sets available in volume
`
`Table 4: Comparison of DMT and CAP
`
`It should be pointed out that DMT won a major victory when the ADSL Joint Procurement Consortium, a
`collection of four RBOCs (Regional Bell Operating Companies), elected to go with a system using both
`ADSL-DMT and ATM.
`
`RELATED ARTICLES
`
`EDN Access -- 03.14.96
`Delivering digital vide <
`https://www.edn.com/design/cons
`ttps://www.edn.com/design/consumer/435017
`By EDN Staff, 14.03.96
`access-
`-03-14-96-
`
`< h
`
`Modulation Schemes: Moving
`Digital Data With Analog Signals
`<
`ttps://www.edn.com/electronics-
`By EDN, 04.10.97
`news/4196988/modulation-
`schemes-
`
`< h
`
`
`
`https://www.eetimes.com/document.asp?doc_id=1255090
`
`11/11
`
`
`
`00011
`
`