telecom@eecs.nwu.edu (TELECOM Moderator) (09/30/90)
TELECOM Digest Sat, 29 Sep 90 18:45:00 CDT Special: ISDN Introduction
Inside This Issue: Moderator: Patrick A. Townson
An Introduction to ISDN From the CERFnet News [Excerpted by Jody Kravitz]
----------------------------------------------------------------------
From: Jody Kravitz <foxtail!kravitz@ucsd.edu>
Subject: An Introduction to ISDN From the CERFnet News
Date: Sat 29 Sep 90 18:00:00 CDT
The most recent issue of the CERFnet news contained a long and useful
article on ISDN. I've excerpted the article from the newsletter:
CERFnet News
California Education and Research Federation Network
August-September 1990
Volume 2, Number 5
<introduction and 5 articles deleted>
AN INTRODUCTION TO ISDN
by Dory Leifer
Motivated by the ever increasing public need to send digital
information in the form of voice, data or image, national governments
along with private corporations have developed a scheme called
Integrated Services Digital Network (ISDN). Although this concept
dates back to the early 1970s, only recently have standards been
developed. The standardization of ISDN has resulted in an emerging
market of ISDN equipment and service plans. This technology will have
widespread impact on both suppliers and users of network equipment and
services.
In the United States, all seven regional Bell operating
companies have initiated limited testing and deployment of ISDN.
General deployment is expected during the mid to late 1990s. Our
European and Japanese counterparts are committed to the nationwide
implementation of ISDN.
This article introduces the basic concepts of telephone
networks and ISDN and explores possible applications of ISDN
technology.
The telephone network
In order to understand why ISDN evolved, let's look at the
current telephone network. The basic telephone is an analog instrument
connected to a pair of wires. The pair of wires from a subscriber's
premises, a private home for example, is connected over approximately
a mile of cable to a local telephone company's central office. This
pair of wires is commonly called the "last mile" or local loop.
Inside the central office, the pair is attached to a device
called a switch. The switch converts the analog signal to digital by
sampling it thousands of times a second. The switch also routes the
call by examining the telephone number called. If the call is
long-distance, it is routed by the local telephone company, Michigan
Bell, for example, to an Interexchange Carrier (IEC) such as AT&T,
MCI, or US Sprint. The IEC routes the call to the local telephone
company at the destination, still preserving the digital nature of the
signal.
This conversion between analog and digital seems reasonable
for voice since humans (even programmers) cannot hear or speak
digitally. But what if we intend to exchange digital information by
connecting two computers together? In that case, we must convert
digital information from our computers into analog signals using a
modem.
When these signals reach the central office, they are
converted back to digital. The reverse process is used at the
destination switch to convert the digital signal back to analog and
pass it to the destination modem which finally turns it back for the
last time to a computer bit stream.
This process is not only redundant, it is inefficient. When
voice is converted from analog to digital, a bit rate of 56,000
bits-per-second (bps) is typically dedicated to carrying it. This rate
is required to make sure that the voice will sound natural when it is
converted back to analog. Since the telephone network treats modems
the same way, a rate of 56,000 bps is also required to convey modem
signals. However, most modems send and receive at or under 2400 bps.
The rest of the capacity is wasted.
Modems serve another purpose apart from digital transmission.
Most modern modems incorporate automatic dialing and answer functions.
We say that an autodial modem exchanges signalling information with
the telephone network. The modem can be instructed to place a call
and report its progress: examples of what it can report back are
"ringing", "busy", and "no circuits available".
Again in this case, because the telephone network is designed
for voice, computer equipment is disadvantaged. The modem requires
special hardware to detect (actually to listen and guess) the sound of
a busy signal, ring, or call incomplete message (usually preceded by
three tones). This type of signalling is not only analog but it is in
band: that is, signals and real transmitted information use the same
channel. Sharing a single circuit to convey both transmission and
signalling information imposes serious limitations.
ISDN relieves the limitations of both in-band signaling and
analog transmission. The next section describes a standard ISDN
interface which provides end-to-end digital transmission and separates
the signaling functions from the transmission functions. ISDN basic
rate interface.
The ISDN basic rate interface is the standard interface to
connect subscribers to the ISDN. This interface uses the existing
telephone wire pair. Instead of using this pair for analog signaling
and transmission, only digital information is conveyed. On this wire,
three channels or digital paths exist. The channels are multiplexed
by giving each a time slice on the wire. Since ISDN channels are half
duplex or uni-directional, a "ping-pong" method is used so that when
one end transmits, the other listens. The ping pong happens with every
tick of some central clock so the link appears to be bidirectional.
Each ISDN circuit includes three channels:
* 2 B or Bearer channels for data or voice (each
64,000 bps)
* 1 D or Data channel for signaling or packet
data (16,000 bps)
These channels provide both signaling and transmission. Notice that
there is no distinction between voice and data on the B-channel. The
ISDN treats both as a stream of bits. The bits have significance only
to the terminating equipment such as a telephone for voice or a
computer for data. When a subscriber wishes to place a call, the
terminating equipment sends a packet on the D-channel containing the
information needed by the network in order to establish the call.
Assuming that the call succeeds, the subscriber may then send either
voice or data on a B-channel. To end the call, a take-down packet is
sent. This is analogous to hanging up.
Bearer channel transmission
The B-channel is referred to as a clear channel because of its
ability to pass an arbitrary bit stream transparently. In reality,
arbitrary bit patterns have limited uses since the B-channel must
adhere to the disciplines of existing voice and data networks. Sending
voice using some non-standard encoding would preclude placing calls
between the ISDN and the existing telephone network. A standard Pulse
Code Modulation (PCM) scheme has been standardized for digitized voice
because it is compatible with the existing voice network.
Correspondingly, a data protocol must be employed on the
B-channel if the subscriber is to reach hosts on the existing packet
services which are not yet on the ISDN. Even if the host is on the
ISDN, the network provides no guarantee that the data will be
transmitted without errors. This is not a serious problem with
terminal sessions (we live with error-prone modems), but for computer
to computer connections (for example, performing a file transfer) an
error-correction protocol may be required.
The B-channel itself provides services that comply with layer
one of the Open Systems Interconnection (OSI) Reference model (the
physical layer). That is, it offers a medium through which bits may
pass.
If a subscriber uses the ISDN to call another computer
directly, a minimum of a layer-two protocol is involved for error
correction and flow control. In many cases, the subscriber will wish
to access a host on a packet network like Telenet. In this case, both
a link layer (OSI layer two) and network layer (layer three) are
required. The subscriber then uses the X.25 protocol between the ISDN
and his or her machine. An interworking unit acts as a gateway between
the ISDN and the packet network, using the X.75 protocol.
A somewhat similar service could be deployed by Merit in the
future to provide Internet access for ISDN subscribers. Off-campus
users could place an ISDN call to an Internet gateway. They could then
access TCP/IP applications like file transfer, remote terminal, and
mail. ISDN provides added support in this case: since the ISDN would
report the caller's address, a unique Internet address could be
associated with a particular calling address. Other services which
require authentication of the caller would also be facilitated by this
feature.
The data channel
The Data or D-Channel was originally specified by the CCITT
for signaling but later was re-specified to include both signaling and
transmission of packet data. Unlike its sister B-channel, the
D-channel is not designed to carry an arbitrary bit stream. The
D-channel uses both a link layer, Link Access Protocol-D (LAPD),
similar to HDLC, and a network layer, Q.931, similar to X.25.
The D-channel may be used for packet data when data throughput
is not of high priority. No call set-up or take-down is required when
using the D-channel to interface in packet mode.
The signaling protocol on the D-channel is based on the set of
signaling messages needed to establish and release a simple 64,000 bps
B-channel voice or data connection. Included in call set-up are:
* Flexible addressing compatible with many standard
network
* Required data rate
* IEC (long distance carrier) selection
if applicable
* Notification if line forwarded to
another address
* User information text
Signaling information is exchanged between a subscriber and the ISDN.
But this information must also be passed within the ISDN to assure
timely circuit establishment, efficient allocation of resources, and
accurate billing and accounting between various service providers. A
protocol called Common Channel Signaling Number Seven (CCS7) performs
these functions. CCS7 was designed by AT&T and is based on the
international standard CCITT Signaling System Seven (SS7). CCS7 is
already used on a wide scale for signaling in the non-ISDN world but
will be essential to support ISDN.
Equipment
Compatibility with existing equipment is extremely important
to most of the users who will migrate from switched and private
networks to ISDN. Therefore, most of the early ISDN equipment which
users will purchase will be adapters for non-ISDN devices such as
asynchronous terminals with RS-232 interfaces, 3270 style terminals
with IBM SDLC and coax interfaces, and various LANs. An interface to
connect common analog telephones will surely be a hot seller.
Many of these devices are quite complex because they have to
support both signalling and transmission. For example, an adapter
which allows RS-232 attachment for terminals needs to interface with
both the B- and D- channels.
Under development by several manufacturers are integrated
terminals that combine voice, data, and signaling into a compact
desktop package. Initially, these terminals will function as
expensive desktop space savers, replacing a separate phone and
terminal, but later they will provide access to truly integrated
services.
What is an integrated service?
An integrated service is one that is capable of providing a
wide assortment of information well organized into a single package.
This information may be, for example, in the form of voice, computer
data, video, or facsimile.
Initially, services available on ISDN will not be integrated.
Voice and data, although they may be accessed together on an
integrated terminal, have little to do with one another. Voice calls
will involve only voice and data calls only data. We speak of this
relationship as Service Coexistence.
The second generation of ISDN services will be integrated. For
example, consider a future bank credit card service. A card holder who
disputes an entry in the credit card bill places an ISDN call to the
bank. At the bank, a customer representative equipped with an ISDN
terminal answers the call. The bank representative immediately has
access to the caller's name and records since the ISDN passes the
customer's originating address. The bank uses this address as a key
into its customer database. The representative can address the
customer by name when answering the phone. When the customer explains
the nature of the problem, the bank representative retrieves the
previous month's bill, which appears simultaneously on both screens.
If the statement is in error, the balance can be recomputed before the
customer's eyes. Integrated services can also facilitate research
collaboration via multi-media voice, image, and control functions
between scientists.
Applications which require exchange of only short, infrequent
messages can use services offered by the D- channel. Applications such
as burglary alerting, energy control, credit card verification, cable
TV requests for service, and home shopping can be accomplished using
the D-channel packet facilities.
Advantages of circuit switching
Although the data rate of 64,000 bps may be too slow for
bandwidth-intensive applications like real-time high definition
imaging, ISDN's circuit-switched capabilities do offer several
advantages to the research community over packet-switched networks
like Merit, NSFNET or ARPANET. Certain real-time applications which
require cross-country connectivity can be run over ISDN. Although the
individual circuits which comprise modern packet networks may be much
faster than 64,000 bps, the overhead involved in packet switching and
queueing is far in excess of similar circuit switching functions on an
established call.
Packet networks try to optimize aggregate performance across
the entire network. Real-time applications are usually interested not
in averages but rather in worst cases. If you get a 64,000 bps ISDN
circuit, you will be guaranteed 64,000 bps service for the duration of
the connection. Throughput on a packet network might average 150,000
bps, for example, but might fall below 64,000 bps 10% of the time,
causing serious problems for a real-time system.
Another advantage ISDN has over packet networks is its
potential ability to interface to a wide variety of digital laboratory
equipment. The ISDN B-channel offers clear channel transmission. There
is no protocol overhead involved in order to exchange information.
This bit pipe can be used, for example, between detector/collector
paired devices without the complication and expense of packet protocol
gateway machines at each end of the connection. ISDN interfaces will
eventually be readily available in VLSI, which will allow them to work
with a wide variety of equipment at minimal additional cost.
High speed (broadband) ISDN
Many argue that 64,000 bps, based on the transmission capacity
of the existing telephone system, is too slow to provide a wide
assortment of integrated services. High-definition television,
computer-aided design, medical imaging, and high-quality audio all
require far more bandwidth than available in the current ISDN. An
evolving standard for broadband ISDN (B-ISDN) may include 150
Megabit-per-second subscriber lines over fiber optic local loops.
Conclusion
ISDN will extend the capabilities of today's telephone
networks, thus providing a market for new services. Most introductory
services will apply service co-existence; services will be described
as "running over" ISDN. ISDN will do for data networks what the
Communications Act of 1934 did for voice -- provide a ubiquitous
method for public transmission. Pioneer users of this technology will
have both the opportunity and the challenge of helping to shape the
future of telecommunications. *
(Dory Leifer is a programmer for the Merit Computer Network, located
in Michigan. This article was originally published in the Merit
Network News, Vol 3 # 3, October, 1988).
--------------
CERFNET NEWS AVAILABLE IN HARD COPY
Send a request to help@cerf.net if you would like to be added to the
hard copy distribution of CERFnet News. Postscript versions are also
available via anonymous ftp to NIC.CERF.NET in the subdirectory
cerfnet_news.
------------------------------
End of TELECOM Digest Special: ISDN Introduction
******************************goldstein@delni.enet.dec.com (Fred R. Goldstein) (10/04/90)
Dory Leifer has done some really good stuff getting TCP/IP running over ISDN. I'd just like to clarify some of the terminology and other details from his CERFnet newsletter article. >This process is not only redundant, it is inefficient. When >voice is converted from analog to digital, a bit rate of 56,000 >bits-per-second (bps) is typically dedicated to carrying it. This rate >is required to make sure that the voice will sound natural when it is >converted back to analog. Since the telephone network treats modems >the same way, a rate of 56,000 bps is also required to convey modem >signals. However, most modems send and receive at or under 2400 bps. >The rest of the capacity is wasted. Actually, the network assigns 64,000 bps per voice call. Since the existing network isn't perfectly bit-transparent, the network usually sets one bit out of eight and gives data users a 56,000 bps clear-channel pipe. You can also sometimes get a 64,000 bps pipe and take your chances, if your telco's willing. >(ISDN Basic Rate Interface...) On this wire, >three channels or digital paths exist. The channels are multiplexed >by giving each a time slice on the wire. Since ISDN channels are half >duplex or uni-directional, a "ping-pong" method is used so that when >one end transmits, the other listens. The ping pong happens with every >tick of some central clock so the link appears to be bidirectional. Actually, the local loop "U" reference point is not ping-pong, but full duplex using echo cancellation. That is, everybody listens and talks on the same wire all the time, cancelling out what you send to pick out what the other guy is sending. This takes modestly heavy silicon but the chips are now out there. Ping-pong was discarded a few years ago, though it's found in some proprietary vendor equipment. At the inside-building "S/T" reference point, they use four wires. >* 1 D or Data channel for signaling or packet People often ask what the "B" and "D" stand for. B stands for Bearer, though H channels are also bearers ("High capacity"). D, however, formally stands for "D". Long-time ISDN weenies may remember that early on, some people discussed how that channel was used for making and breaking calls, thus causing change (delta) to the B channels. But it's not a delta channel any more, just a D. Right. You Will Forget Delta. (It was never, however, Data; most data flows on B channels.) >These channels provide both signaling and transmission. Notice that >there is no distinction between voice and data on the B-channel. The >ISDN treats both as a stream of bits. Not exactly. If you ask for a stream of bits, you get it. But if you ask for "speech" or "audio", the network has the right to process your bits as it desires, preserving the audio content. If you call between North America and Europe, the network MUST change speech and PCM audio because Europe and North America use different PCM standards! They're mutually unintelligible, though both are 64 kbps PCM. Similarly, the network MUST NOT change a clear channel (data). >No call set-up or take-down is required when using the D-channel to >interface in packet mode. Not exactly that simple. If you use the B channel, you have to first set up a circuit call to the packet handler, THEN set up an X.25 call. If you use the D channel, you still have to set up the X.25 call, but using DSS1 (Q.931) instead of X.25 for the call establishment phase. It's one step instead of two, but not connectionless. (Yet. Just not enough datagram fans in CCITT.) Fred R. Goldstein Digital Equipment Corp., Littleton MA goldstein@delni.enet.dec.com voice: +1 508 486 7388 Do you think anyone else on the planet would share my opinions, let alone a multi-billion dollar corporation?
kevinc@uunet.uu.net (Kevin Collins) (10/04/90)
In-Reply-To: article <12768@accuvax.nwu.edu>, by Dory Leifer, sent in by Jody Kravitz: The article was very informative, but it left out a few items that I spotted, so here goes ... First of all, the article mentioned Basic Rate Interface (BRI), which is the local connection between the CO (or PBX) and the end user. It did not mention, however, Primary Rate Interface (PRI), which is the LD connection between CO's. PRI (in the US) is 23 data (B) channels and 1 control (D) channel, with all channels, both B and D, running at 64Kb/second. PRI and BRI use the same call control messages (Q.931). [Article talks about Europeans and Japanese implementing ISDN standard] Yes, they are implementing the standard, but they are doing some things differently than US manufacturers. Case in point: ISDN PRI here is 23B+D, in Europe, China, and Japan(?) it's 30B+2D. There are even differences between MCI's implementation of PRI and AT&T's implementation. Oh well, typical standard :-). [Future bank credit card service example, service rep gets customer's info from calling number, bill appears on both the rep's screen and the customer's screen, etc.] The part concerning the service rep being able to access the customer's data from the calling number is possible NOW, with PRI ANI and a link between the bank's ACD and their computer. Both BRI and PRI would be needed for the entire example to work. The customer's billing data would go over BRI from the bank to the bank's CO, over PRI from CO to CO, and over BRI from the customer's CO to the customer's BRI device, where it would be displayed. [Lack of current broadband standard] The current version of the standard has provisions for using 6 PRI B-channels together (called an H0 channel, 384 Kb/sec) and using 24 B-channels together (H11 channel, 1.536 Mb/sec [this is AT&T's number, don't know why it's not 1.544Mb/sec]). AT&T offers a "Switched 384" service (the H0 channel), but I don't think they offer the H11 channel service yet. I don't know what services of this nature MCI or Sprint offers. Kevin Collins Aspect Telecommunications USENET: ...uunet!aspect!kevinc San Jose, CA ------------------------------ From: Julian Macassey <julian@bongo.uucp> Subject: Re: Question About "Point of Demarcation" Date: 3 Oct 90 17:19:22 GMT Organization: The Hole in the Wall Hollywood California U.S.A. In article <12849@accuvax.nwu.edu>, dgc@math.ucla.edu (David G. Cantor) writes: > In TELECOM Digest, V10, No. 693, Roger Clark refers to new FCC > regulations concerning inside wirng rules and, in particular, refers > to "the point of demarcation" between the telco's wiring and the > subscriber's wiring. > Does the FCC require that there be such a point of demarcation? I > live in GTE country and neither I, nor my neighbors, have such a > point. Does this point (which I assume is a modular jack and plug) > have to be accessible without entering the subscriber's premises, or > at least without passing through a locked gate or door? You, your neighbours and everyone in Southern California have had such a point for some years. In fact you may be paying $0.50 a month or so for "free" maintenance of your inside wiring - check your phone bill. This wiring you are paying to have maintained starts at your demarc'. The demarcation point which is the physical location where the telco responsibility for wire ends and yours begins. This is similar to the electric meter, everything after it is your wire and everything before it is Edison's wire. You can mess with your wire, you can't mess with Edison's wire. But, yes they meter your calls at the CO, not at your demarc point. The demarc is not always easy to get to. Especially if in a basement which is usually locked. To save the subscriber the grief of having to be home when the telco drops by, it is convenient to have an accessible demarc, but not essential. So where is your demarc point? In Southern California, it is usually in a little box on the wall of a house with an aerial wire leading to it. Inside the box is a device called a protector, it looks like two nuts and a ground wire, this is where the house wire connects to the telco "drop wire". Some houses have the demarc in the crawl space - many of these in West LA, Beverly Hills. Modern Demarc's are called Network Interfaces and besides the protectors they also have an RJ11 jack so that you can separate house wire from the drop so you can plug a phone in there to determine if your wire is bad or the telco circuit is bad. Apartment houses usually have all the demarcs in an easily accessible closet. Finding the right one for an apartment can sometimes be a challenge. Office buildings usually have them in the "telco closet", at least one on each floor. The demarcs are usually orange covered punch down blocks with the subscribers name on them, they are loosely referred to as the "RJ-21X". In those parts of the US that have real basements, the residential demarc is usually in the basement. Julian Macassey, n6are julian@bongo.info.com ucla-an!denwa!bongo!julian N6ARE@K6IYK (Packet Radio) n6are.ampr.org [44.16.0.81] voice (213) 653-4495 ------------------------------ Date: Wed, 3 Oct 90 13:20:45 edt From: Bob Goudreau <goudreau@dg-rtp.dg.com> Subject: Re: Which Came First? Reply-To: goudreau@dg-rtp.dg.com (Bob Goudreau) Organization: Data General Corporation, Research Triangle Park, NC In article <12879@accuvax.nwu.edu>, jwb@monu6.cc.monash.edu.au (Jim Breen) writes: > As I have heard it, the ISO standard for numeric keypads antedated the > CCITT recommendation. When CCITT "studied" the keypad layout, AT&T > representatives refused point-blank to compromise, and CCITT > (cravenly) gave in. Hmmm. So making AT&T switch to the 7-8-9 layout would have been mere "compromise", but having CCITT adopt the 1-2-3 layout (which was found to be superior in human factors experiments run by both AT&T and CCITT) was "cravenly" giving in. Odd, that. > All praise to those (few) PTTs which held out and adopted the ISO > version. Au contraire; all praise to those (many) PTTs which adopted the CCITT recommendation. Standardizing inferiority is certainly not progress. Bob Goudreau +1 919 248 6231 Data General Corporation 62 Alexander Drive goudreau@dg-rtp.dg.com Research Triangle Park, NC 27709 ...!mcnc!rti!xyzzy!goudreau USA ------------------------------ From: Rop Gonggrijp <ropg@ooc.uva.nl> Subject: Re: Equivalents of 800/900/976/911 Numbers in the Netherlands Date: 3 Oct 90 18:18:39 GMT Organization: uvabick hansm@cs.kun.nl (Hans Mulder) writes: >In the Netherlands PTT Telecom has managed to confuse everybody by >creating a single new area code (06) containing the equivalents of >both 800 and 900 numbers. The Consumers' Association has demanded >that the toll numbers be changed into 07 numbers, but since all 07X >area codes are already in use, this is not possible. >There is no 06-[3589] blocking for residential customers. They do >provide 06-blocking for PBXs. This also blocks 06-11. No no, not true: on one of my residential phone-lines I have outgoing call blocking ONLY for 069 and 063. They even offer this service on old stepper-switches (by adding a piece of hardware between the switch and your wires). >Next time I'll tell you about the night when PTT Telecom intended to >demonstrate that 06 was also usable as the choke exchange and found >out the hard way that it was not. Oh yeah, I had fun that night waiting for a dialtone for up to twenty minutes! By the way: a lot of the numbers in the FREE series 06-022XXXX end up outside of Holland (like 06-0229111 for AT&T USADirect) and some of them route over lines with IN-BAND signalling systems. Have PHUN! Rop Gonggrijp (ropg@ooc.uva.nl) is also editor of Hack-Tic (hack/phreak mag.) Postbus 22953 (in DUTCH) 1100 DL AMSTERDAM tel: +31 20 6001480
meier@uunet.uu.net (Rolf Meier) (10/04/90)
In article <12978@accuvax.nwu.edu> goldstein@delni.enet.dec.com (Fred R. Goldstein) writes: >bits as it desires, preserving the audio content. If you call between >North America and Europe, the network MUST change speech and PCM audio >because Europe and North America use different PCM standards! They're >mutually unintelligible, though both are 64 kbps PCM. Similarly, the >network MUST NOT change a clear channel (data). Actually, if you decoded ulaw with an Alaw decoder, or vice versa, the difference is practically inaudible compared to the use of the proper decoder. However, the conversion is made anyway, in order to meet the quantization requirements. Rolf Meier Mitel Corporation
BRUCE@ccavax.camb.com (Barton F. Bruce) (10/06/90)
In article <12979@accuvax.nwu.edu>, aspect!kevinc@uunet.uu.net (Kevin Collins) writes: > The current version of the standard has provisions for using 6 PRI > B-channels together (called an H0 channel, 384 Kb/sec) and using 24 > B-channels together (H11 channel, 1.536 Mb/sec [this is AT&T's number, > don't know why it's not 1.544Mb/sec]). AT&T offers a "Switched 384" The 1.536 Mb/sec is 64kb x 24. The oft used 1.544 figure includes the additional 8kb for the framing bit. Each 1/8000 of a second, the line passes 192 bits of data (24 x 8) + one framing bit for a total of 193 bits. Other than keeping frames in sync and defining the A and B signaling frames (or more generally, where one is within the super-frame format), the framing bit also (under ESF) can carry a small amount of network managment data.
elliott@uunet.uu.net (Paul Elliott x225) (10/09/90)
In article <13051@accuvax.nwu.edu>, mitel!spock!meier@uunet.uu.net (Rolf Meier) writes: > In article <12978@accuvax.nwu.edu> goldstein@delni.enet.dec.com (Fred > R. Goldstein) writes: > >bits as it desires, preserving the audio content. If you call between > >North America and Europe, the network MUST change speech and PCM audio > >because Europe and North America use different PCM standards! They're > >mutually unintelligible, though both are 64 kbps PCM. Similarly, the > >network MUST NOT change a clear channel (data). > Actually, if you decoded ulaw with an Alaw decoder, or vice versa, the > difference is practically inaudible compared to the use of the proper > decoder. However, the conversion is made anyway, in order to meet the > quantization requirements. While it is true that the mu-law and A-law encoding/decoding curves are very similar, the actual digital representation of the signals is quite different, requiring code conversion to be intelligible. Mu-law and A-law codecs both use a quasi-logarithmic transfer function, to obtain optimal signal-to-noise ratios over a wide dynamic range. The quasi-log characteristic is achieved by breaking a non-linear transfer function into a series of linear "chords", with each chord consisting of several equal-sized steps. The step size is doubled for each successive chord (the piecewise approximated curve is symmetrical about zero). Thus, for a given full-scale value, signals closer to zero are encoded with greater precision than would be obtained with a linear code. The resulting encoding gives nearly equal stepsize (when measured in dB) for signals within the encoding range. The dynamic range of the mu-law codec is approximately 72 dB, which compares well to the 42 dB range of a linear 8-bit code (seven bits plus sign). The mu-law function provides eight chords, of 16 steps each, while for some reason, the European A-law standard has a first chord of 32 steps, and six remaining chords of 16 steps. Mu-law provides better S/N over the full range, while A-law gives reduced distortion at low levels. These differences are almost inaudible, but the standards threw in a big monkey wrench. Mu-law encoding could be called "bit-inverted sign-magnitude", where "positive full scale"= 10000000 "positive zero" = 11111111 "negative zero" = 01111111 "negative full scale"= 00000000 A-law inverts alternate bits, to give: "positive full scale"= 10101010 "positive zero" = 11010101 "negative zero" = 01010101 "negative full scale"= 00101010 I guarantee you, this WILL be noticed! Still, as far as standards are concerned, I guess we "telecom types" don't have it as bad as some other technical fields... Paul M. Elliott Optilink Corporation (707) 795-9444 {uunet, pyramid, tekbspa}!optilink!elliott