<no-top-post> wrote in message news:1161962862.317275@work.isdsl.net...> It's common knowledge that digital technology gives more telephone > [4 Khz wide] channels than analog technology - for the same spectrum > width. This even with the steep-sided filters and SSB technology. > > Since the input and output need to be in analog form, I find > digital's claim very unintuitive -- almost doubting it.This can be a difficult subject. One place to start is to investigate quadrature amplitude modulation (QUAM) and learn how several binary numbers can be encoded into each chunk of a transmission stream.
How can digital be more spectrum efficient than analog ?
Started by ●October 27, 2006
Reply by ●October 27, 20062006-10-27
Reply by ●October 27, 20062006-10-27
Charles Schuler wrote:> <no-top-post> wrote in message news:1161962862.317275@work.isdsl.net... >> It's common knowledge that digital technology gives more telephone >> [4 Khz wide] channels than analog technology - for the same spectrum >> width. This even with the steep-sided filters and SSB technology. >> >> Since the input and output need to be in analog form, I find >> digital's claim very unintuitive -- almost doubting it. > > This can be a difficult subject. One place to start is to investigate > quadrature amplitude modulation (QUAM) and learn how several binary numbers > can be encoded into each chunk of a transmission stream.Or look at the required SNR fro audio and the required SNR for a GMSK or BPSK signal. And then take into account the influence of neigbouring cells. Thomas
Reply by ●October 27, 20062006-10-27
Charles Schuler wrote:> <no-top-post> wrote in message news:1161962862.317275@work.isdsl.net... >> It's common knowledge that digital technology gives more telephone >> [4 Khz wide] channels than analog technology - for the same spectrum >> width. This even with the steep-sided filters and SSB technology. >> >> Since the input and output need to be in analog form, I find >> digital's claim very unintuitive -- almost doubting it. > > This can be a difficult subject. One place to start is to investigate > quadrature amplitude modulation (QUAM) and learn how several binary numbers > can be encoded into each chunk of a transmission stream.QAM can be entirely analog. Digital signals are more tolerant of moderate amounts of noise, but that's another (long) story. Jerry -- "The rights of the best of men are secured only as the rights of the vilest and most abhorrent are protected." - Chief Justice Charles Evans Hughes, 1927 ���������������������������������������������������������������������
Reply by ●October 28, 20062006-10-28
On Fri, 27 Oct 2006 10:29:06 -0500, no-top-post wrote:>It's common knowledge that digital technology gives more telephone >[4 Khz wide] channels than analog technology - for the same spectrum >width. This even with the steep-sided filters and SSB technology. >T'aint nesessarily so. As another poster pointed out FDM could squeeze 24 channels os SSB 4kz voice into 96 KHz. where a T1 takes over 1.5 megabits/sec to do the same thing.>Since the input and output need to be in analog form, I find >digital's claim very unintuitive -- almost doubting it. > >Can someone offer an intuitive explanation ? > >I guess a 4 HHz bandwidth telco signal contains a lot of >redundancy which is not needed to convey the inteligence ? >True of any voice transmission. Unless both parties are in the habit of talking at the same time 50% of the "bandwith" (I hate that term but everbody thinks they know what it is :< ) is wasted. Some forms of digital transmission, i.e. VOIP claim to be able to reclaim that waste.>Speech can even be transmitted as a sequence of 'predetermined >sylables' - what are they called again ? >Political Speeches ?!? Telcos sometimes cut the 64 Kbit/VC rate down to 32 Kbits for a 2-1 improvent in transmission capacity, This doesn't really sound any worse for voice but don't try to send high speed data over it :(. CODECS used in cellphones take this approach- they divide the voice bandwith into upper & lower bands. They pretty much encode the lower frequencies directly. These lower freqs are what gives everybody thier distinctive voice, or "timber". As the freqs are low they don't take much bit rate to transmit. The upper frequencies, which are what actually convey the "intelligence", or, if calling 1-800-PSYCHIC "lack of intellegence" are analyzied for fricatives, glottal stops & for all I know spitting lisps. Once analyzed the CODEC issues a code representing the particular type of sound it think it just heard. This whole process of encoding can take a half a second or so as the CODEC likes to here what comes next after what you just said as part of the sound recognition process, this is why you get a nasty echo coming back at you if theres bad return loss on the analog line & the echo cancellers at the CO are on a coffee break. Depending on the codec standard you can squeeze a standard 64K telco D4 standard digitized signal down to about 12 K or so. The cellular TDMA IS91 got as low as 8K but sounded like crap & didn't like bit errors one little- er, um, bit. As for how much "real" bandwidth this occupies, well the IS91 originally tried to squeeze 6 8 kbit channels in the the 10Khz RF channel space that previously the analog the AMPS analog FMd the 4kHz voice channel on. They gave up on that & went to three 12+ change kbit channels per . Still sounded like shit, though not as bat as the 8 but on the RF side suffers from adjacent cell & co channel interference to a nasty degree. But a three to one improvement in using the RF space. GSM also uses a 12.5 codec system but puts 8 voice streams in a 200 KHz RF channel. This obviously is not as efficient spectrally as TDMA or even the aforementioned FDM SSB signal, but it sounds great & is far less sensitive to bit errors or interference than TDMA The codec is much improved in how it encodes stuff with error correction & dividing the bit stream into Very Imprtant Bits, Not So Important Bits & Bits We Really Don't Give a Rat's Ass About. Highest level of error correction is dedicated to the the VIBs, some to the NSIBs, & I'll let you guess about the last catagory ;>.>So it seems to me that the improvement over analog methods >can't be acheived merely by 'modeming', which would only add >entropy. But once it's in the digital format, standard compression >[by removing redundancy] can make for more efficiency ?Now, as I mentioned earlier, some transmission standards claim to give maximum efficiency by filling in the gaps in the data stream with other data, whether it by 911 calls or on-line porno. I have a sneaky suspician that by the time you look at all the overhead on these systems with the data getting packetized & constantly re-packaged as it moves through different levels of the system that the actual savings in "bandwidth" (theres that *%&$ term again :( ) arn't all that great. If it routed through a virtual circuit on a SONET stream I don't believe that the "50%" "wasted bandwidth" (GRRRR) is even available to other data streams 'though I may be wrong on that :/. I'd also point out that that "bandwidth", at least on the terrestrial side, is dirt cheap. (Do you think maybe 'cause thats what fiber is made outta???) So why use it? FDM was a pain in the tucus to keep lined up. Every step it went through in the transmission path added noise- even on a good system long distance calls sounded like long distance calls. Digital, at least uncompressed digital on a stable path, sounds the same no matter how far you send it. Advances in high speed digital circuitry allows the equipment used to handle the data streams to take up way less space than the older analog equivalents & for a fraction of the price>Thanks for any info,You're welcome :).> >== Chris Glur. >H.
Reply by ●October 28, 20062006-10-28
Howard Eisenhauer <howarde@REMOVECAPShfx.eastlink.ca> wrote:>On Fri, 27 Oct 2006 10:29:06 -0500, no-top-post wrote: > >>It's common knowledge that digital technology gives more telephone >>[4 Khz wide] channels than analog technology - for the same spectrum >>width. This even with the steep-sided filters and SSB technology. >> >T'aint nesessarily so. As another poster pointed out FDM could >squeeze 24 channels os SSB 4kz voice into 96 KHz. where a T1 takes >over 1.5 megabits/sec to do the same thing.Comparing bandwidth to bit rate is nonsense! 96 KHz does not compare to 1.5 Mbps in any productive way. Keep in mind that a telco voice channel is 4 KHz analog, but over the same bandwidth one can easily send a 64 Kbps digital PAM signal (i.e., v.90) using the exact same codec in the telephone switch line card. Bandwidth does not use more bitrate, it provides more! Shannon showed channel capacity to be relative to both bandwidth and signal-to-noise ratio. Anything you do to change one *requires* a change in the other to maintain the same maximum capacity. Different carrier systems do exactly that, balancing between SNR and bandwidth. Bandwidth is not what makes digital more effective than analog. Above a minimum SNR digital systems are essentially error free while analog system have additive noise. For that reason digital is usually prefered. The second most significant reason is the ease with which digital carrier systems can be designed to trade SNR for bandwidth by using multi-level encoding, which means well matched adaptions to anything from very noisy fiber optics to very quiet coax cable. -- Floyd L. Davidson <http://www.apaflo.com/floyd_davidson> Ukpeagvik (Barrow, Alaska) floyd@apaflo.com
Reply by ●October 28, 20062006-10-28
Howard Eisenhauer <howarde@REMOVECAPShfx.eastlink.ca> wrote:>On Fri, 27 Oct 2006 10:29:06 -0500, no-top-post wrote: > >>It's common knowledge that digital technology gives more telephone >>[4 Khz wide] channels than analog technology - for the same spectrum >>width. This even with the steep-sided filters and SSB technology. >> >T'aint nesessarily so. As another poster pointed out FDM could >squeeze 24 channels os SSB 4kz voice into 96 KHz. where a T1 takes >over 1.5 megabits/sec to do the same thing.Comparing bandwidth to bit rate is nonsense! 96 KHz does not compare to 1.5 Mbps in any productive way. Keep in mind that a telco voice channel is 4 KHz analog, but over the same bandwidth one can easily send a 64 Kbps digital PAM signal (i.e., v.90) using the exact same codec in the telephone switch line card. Bandwidth does not use more bitrate, it provides more! Shannon showed channel capacity to be relative to both bandwidth and signal-to-noise ratio. Anything you do to change one *requires* a change in the other to maintain the same maximum capacity. Different carrier systems do exactly that, balancing between SNR and bandwidth. Bandwidth is not what makes digital more effective than analog. Above a minimum SNR digital systems are essentially error free while analog system have additive noise. For that reason digital is usually prefered. The second most significant reason is the ease with which digital carrier systems can be designed to trade SNR for bandwidth by using multi-level encoding, which means well matched adaptions to anything from very noisy fiber optics to very quiet coax cable. -- Floyd L. Davidson <http://www.apaflo.com/floyd_davidson> Ukpeagvik (Barrow, Alaska) floyd@apaflo.com
Reply by ●October 28, 20062006-10-28
Howard Eisenhauer <howarde@REMOVECAPShfx.eastlink.ca> wrote:>On Fri, 27 Oct 2006 10:29:06 -0500, no-top-post wrote: > >>It's common knowledge that digital technology gives more telephone >>[4 Khz wide] channels than analog technology - for the same spectrum >>width. This even with the steep-sided filters and SSB technology. >> >T'aint nesessarily so. As another poster pointed out FDM could >squeeze 24 channels os SSB 4kz voice into 96 KHz. where a T1 takes >over 1.5 megabits/sec to do the same thing.Comparing bandwidth to bit rate is nonsense! 96 KHz does not compare to 1.5 Mbps in any productive way. Keep in mind that a telco voice channel is 4 KHz analog, but over the same bandwidth one can easily send a 64 Kbps digital PAM signal (i.e., v.90) using the exact same codec in the telephone switch line card. Bandwidth does not use more bitrate, it provides more! Shannon showed channel capacity to be relative to both bandwidth and signal-to-noise ratio. Anything you do to change one *requires* a change in the other to maintain the same maximum capacity. Different carrier systems do exactly that, balancing between SNR and bandwidth. Bandwidth is not what makes digital more effective than analog. Above a minimum SNR digital systems are essentially error free while analog system have additive noise. For that reason digital is usually prefered. The second most significant reason is the ease with which digital carrier systems can be designed to trade SNR for bandwidth by using multi-level encoding, which means well matched adaptions to anything from very noisy fiber optics to very quiet coax cable. -- Floyd L. Davidson <http://www.apaflo.com/floyd_davidson> Ukpeagvik (Barrow, Alaska) floyd@apaflo.com
Reply by ●October 28, 20062006-10-28
Howard Eisenhauer <howarde@REMOVECAPShfx.eastlink.ca> wrote:>On Fri, 27 Oct 2006 10:29:06 -0500, no-top-post wrote: > >>It's common knowledge that digital technology gives more telephone >>[4 Khz wide] channels than analog technology - for the same spectrum >>width. This even with the steep-sided filters and SSB technology. >> >T'aint nesessarily so. As another poster pointed out FDM could >squeeze 24 channels os SSB 4kz voice into 96 KHz. where a T1 takes >over 1.5 megabits/sec to do the same thing.Comparing bandwidth to bit rate is nonsense! 96 KHz does not compare to 1.5 Mbps in any productive way. Keep in mind that a telco voice channel is 4 KHz analog, but over the same bandwidth one can easily send a 64 Kbps digital PAM signal (i.e., v.90) using the exact same codec in the telephone switch line card. Bandwidth does not use more bitrate, it provides more! Shannon showed channel capacity to be relative to both bandwidth and signal-to-noise ratio. Anything you do to change one *requires* a change in the other to maintain the same maximum capacity. Different carrier systems do exactly that, balancing between SNR and bandwidth. Bandwidth is not what makes digital more effective than analog. Above a minimum SNR digital systems are essentially error free while analog system have additive noise. For that reason digital is usually prefered. The second most significant reason is the ease with which digital carrier systems can be designed to trade SNR for bandwidth by using multi-level encoding, which means well matched adaptions to anything from very noisy fiber optics to very quiet coax cable. -- Floyd L. Davidson <http://www.apaflo.com/floyd_davidson> Ukpeagvik (Barrow, Alaska) floyd@apaflo.com
Reply by ●October 28, 20062006-10-28
Howard Eisenhauer <howarde@REMOVECAPShfx.eastlink.ca> wrote:>On Fri, 27 Oct 2006 10:29:06 -0500, no-top-post wrote: > >>It's common knowledge that digital technology gives more telephone >>[4 Khz wide] channels than analog technology - for the same spectrum >>width. This even with the steep-sided filters and SSB technology. >> >T'aint nesessarily so. As another poster pointed out FDM could >squeeze 24 channels os SSB 4kz voice into 96 KHz. where a T1 takes >over 1.5 megabits/sec to do the same thing.Comparing bandwidth to bit rate is nonsense! 96 KHz does not compare to 1.5 Mbps in any productive way. Keep in mind that a telco voice channel is 4 KHz analog, but over the same bandwidth one can easily send a 64 Kbps digital PAM signal (i.e., v.90) using the exact same codec in the telephone switch line card. Bandwidth does not use more bitrate, it provides more! Shannon showed channel capacity to be relative to both bandwidth and signal-to-noise ratio. Anything you do to change one *requires* a change in the other to maintain the same maximum capacity. Different carrier systems do exactly that, balancing between SNR and bandwidth. Bandwidth is not what makes digital more effective than analog. Above a minimum SNR digital systems are essentially error free while analog system have additive noise. For that reason digital is usually prefered. The second most significant reason is the ease with which digital carrier systems can be designed to trade SNR for bandwidth by using multi-level encoding, which means well matched adaptions to anything from very noisy fiber optics to very quiet coax cable. -- Floyd L. Davidson <http://www.apaflo.com/floyd_davidson> Ukpeagvik (Barrow, Alaska) floyd@apaflo.com
Reply by ●October 28, 20062006-10-28
On 10/27/06 8:36 PM, in article kmg5k21ljhik3sgfecg6br34u2llkko0cn@4ax.com, "Howard Eisenhauer" <howarde@REMOVECAPShfx.eastlink.ca> wrote:> On Fri, 27 Oct 2006 10:29:06 -0500, no-top-post wrote: > >> It's common knowledge that digital technology gives more telephone >> [4 Khz wide] channels than analog technology - for the same spectrum >> width. This even with the steep-sided filters and SSB technology. >> > T'aint nesessarily so. As another poster pointed out FDM could > squeeze 24 channels os SSB 4kz voice into 96 KHz. where a T1 takes > over 1.5 megabits/sec to do the same thing. > > >> Since the input and output need to be in analog form, I find >> digital's claim very unintuitive -- almost doubting it. >> >> Can someone offer an intuitive explanation ? >> >> I guess a 4 HHz bandwidth telco signal contains a lot of >> redundancy which is not needed to convey the inteligence ? >> > True of any voice transmission. Unless both parties are in the habit > of talking at the same time 50% of the "bandwith" (I hate that term > but everbody thinks they know what it is :< ) is wasted. Some forms > of digital transmission, i.e. VOIP claim to be able to reclaim that > waste.VOIP, when done at rates lower than 64 kbit/s, permits you to communicate, but many aspects of quality suffer; such as increasing echo with decreasing bit-rate.> >> Speech can even be transmitted as a sequence of 'predetermined >> sylables' - what are they called again ? >> > > Political Speeches ?!? > > Telcos sometimes cut the 64 Kbit/VC rate down to 32 Kbits for a 2-1 > improvent in transmission capacity, This doesn't really sound any > worse for voice but don't try to send high speed data over it :(.Name even one real telco that does that. I can't think of one. ADPCM is used, however, in many private networks.> > CODECS used in cellphones take this approach- they divide the voice > bandwith into upper & lower bands. They pretty much encode the lower > frequencies directly. These lower freqs are what gives everybody > thier distinctive voice, or "timber". As the freqs are low they don't > take much bit rate to transmit. The upper frequencies, which are what > actually convey the "intelligence", or, if calling 1-800-PSYCHIC "lack > of intellegence" are analyzied for fricatives, glottal stops & for all > I know spitting lisps. Once analyzed the CODEC issues a code > representing the particular type of sound it think it just heard. > This whole process of encoding can take a half a second or so as the > CODEC likes to here what comes next after what you just said as part > of the sound recognition process, this is why you get a nasty echo > coming back at you if theres bad return loss on the analog line & the > echo cancellers at the CO are on a coffee break. > > Depending on the codec standard you can squeeze a standard 64K telco > D4 standard digitized signal down to about 12 K or so. The cellular > TDMA IS91 got as low as 8K but sounded like crap & didn't like bit > errors one little- er, um, bit. > > As for how much "real" bandwidth this occupies, well the IS91 > originally tried to squeeze 6 8 kbit channels in the the 10Khz RF > channel space that previously the analog the AMPS analog FMd the 4kHz > voice channel on. They gave up on that & went to three 12+ change > kbit channels per . Still sounded like shit, though not as bat as the > 8 but on the RF side suffers from adjacent cell & co channel > interference to a nasty degree. But a three to one improvement in > using the RF space. > > GSM also uses a 12.5 codec system but puts 8 voice streams in a 200 > KHz RF channel. This obviously is not as efficient spectrally as TDMA > or even the aforementioned FDM SSB signal, but it sounds great & is > far less sensitive to bit errors or interference than TDMA > The codec is much improved in how it encodes stuff with error > correction & dividing the bit stream into Very Imprtant Bits, Not So > Important Bits & Bits We Really Don't Give a Rat's Ass About. Highest > level of error correction is dedicated to the the VIBs, some to the > NSIBs, & I'll let you guess about the last catagory ;>. > > >> So it seems to me that the improvement over analog methods >> can't be acheived merely by 'modeming', which would only add >> entropy. But once it's in the digital format, standard compression >> [by removing redundancy] can make for more efficiency ? > > Now, as I mentioned earlier, some transmission standards claim to give > maximum efficiency by filling in the gaps in the data stream with > other data, whether it by 911 calls or on-line porno. I have a sneaky > suspician that by the time you look at all the overhead on these > systems with the data getting packetized & constantly re-packaged as > it moves through different levels of the system that the actual > savings in "bandwidth" (theres that *%&$ term again :( ) arn't all > that great. If it routed through a virtual circuit on a SONET stream > I don't believe that the "50%" "wasted bandwidth" (GRRRR) is even > available to other data streams 'though I may be wrong on that :/. > > I'd also point out that that "bandwidth", at least on the terrestrial > side, is dirt cheap. (Do you think maybe 'cause thats what fiber is > made outta???) > > So why use it? FDM was a pain in the tucus to keep lined up. Every > step it went through in the transmission path added noise- even on a > good system long distance calls sounded like long distance calls. > Digital, at least uncompressed digital on a stable path, sounds the > same no matter how far you send it. Advances in high speed digital > circuitry allows the equipment used to handle the data streams to take > up way less space than the older analog equivalents & for a fraction > of the price > > >> Thanks for any info, > > You're welcome :). > > >> >> == Chris Glur. >> > > H.
