larry@uunet.uu.net (Larry Lippman) (09/16/90)
In article <11532@accuvax.nwu.edu> dolf@idca.tds.philips.nl (Dolf Grunbauer) writes: > I always assumed that when making a telephone call the line to the > otherside is the same the line back from him to me. The other day > someone told me that this is not the case, especially when making a > international phone call. According to him it is possible that for > example when calling from Europe to the USA one line could use a > satellite connection while the other could use a transatlantic cable. > Is this true? Unless echo suppressors have become *much* more sophisticated than those with which I was once familiar, I would be surprised if such vastly different propagation paths could be used on the E-W and W-E directions of a given intertoll circuit. Control of echo on intertoll circuits can be implemented by simple attenuation in the trivial case, and voice-switched attenuation through echo suppressors in the more common case. The proper design of intertoll circuits, including configuration of echo suppressors, is governed by the Via Net Loss (VNL) concept. VNL design requires knowledge of propagation delay in milliseconds. At the time I was involved in the telephone industry, the VNL design with which I was familiar imposed a maximum of 22.5 milliseconds propagation delay to a DDD switching midpoint, with an maximum overall delay of 45 milliseconds on any given DDD circuit. It was always a "given" that VNL design required the same propagation delay in each direction. While I admit that I have no firsthand experience with intertoll circuits involving satellites or transoceanic cable, I would find it difficult to believe that any satisfactory transmission (and echo) performance could be achieved with the E-W and W-E directions that have widely *differing* propagation times. Obviously, satellite transmission by its very nature imposes propagation delays which far exceed 45 milliseconds. However, echo suppressors and intertoll circuit design can be set up to deal with such increased propagation delays - *provided* that the delay is equal in each direction. Also, I can think of no valid reason to split E-W and W-E routing between different transmission facilities. *ALL* transmission facilities used for intertoll circuits are, by their very nature, bi-directional. While I have seen all sorts of route diversity and failure protection switching, I have never seen anything that split transmission directions through different facilities. Larry Lippman @ Recognition Research Corp. {boulder||decvax||rutgers||watmath}!acsu.buffalo.edu!kitty!larry VOICE: 716/688-1231 || FAX: 716/741-9635 {utzoo||uunet}!/ \aerion!larry
ijk@violin.att.com (Ihor J Kinal) (09/18/90)
Larry Lippman wrote: > Also, I can think of no valid reason to split E-W and W-E > routing between different transmission facilities. *ALL* transmission > facilities used for intertoll circuits are, by their very nature, > bi-directional. While I have seen all sorts of route diversity and > failure protection switching, I have never seen anything that split > transmission directions through different facilities. The way I remember this, the reason that a split was put into effect was excessive delay on satellite calls. Remember that the satellites are at approx 22,000 miles up. Double that, and it means that we need approx a quarter of a second to traverse in ONE direction. [Speed of light = 186,000 m.p.s.]. If both sides went over satellite, that would mean a half-second of extra delay from when one person stopped talking, until the next person could possibly reply. This much delay would then then cause the original person to start talking again, to see if the distant party was still there, leading to great confusion. Ihor Kinal att!cbnewsh!ijk [Include standard disclaimers and although I work at Bell Labs, I never did any work satellite communications - I'm just a software person anyway.]
grayt@uunet.uu.net (Tom Gray) (09/18/90)
In article <12220@accuvax.nwu.edu> kitty!larry@uunet.uu.net (Larry Lippman) writes: X-Telecom-Digest: Volume 10, Issue 651, Message 4 of 9 >In article <11532@accuvax.nwu.edu> dolf@idca.tds.philips.nl (Dolf >Grunbauer) writes: >> I always assumed that when making a telephone call the line to the >> otherside is the same the line back from him to me. The other day >> someone told me that this is not the case, especially when making a >> international phone call. According to him it is possible that for >> example when calling from Europe to the USA one line could use a >> satellite connection while the other could use a transatlantic cable. >> Is this true? >Unless echo suppressors have become *much* more sophisticated >than those with which I was once familiar, I would be surprised if >such vastly different propagation paths could be used on the E-W and >W-E directions of a given intertoll circuit. Echo suppressors have been succeeded by echo cancellers and differing W-E and E-W paths are used in the network. >Control of echo on intertoll circuits can be implemented by >simple attenuation in the trivial case, and voice-switched attenuation >through echo suppressors in the more common case. The proper design >of intertoll circuits, including configuration of echo suppressors, is >governed by the Via Net Loss (VNL) concept. VNL design requires >knowledge of propagation delay in milliseconds. At the time I was >involved in the telephone industry, the VNL design with which I was >familiar imposed a maximum of 22.5 milliseconds propagation delay to a >DDD switching midpoint, with an maximum overall delay of 45 >milliseconds on any given DDD circuit. It was always a "given" that >VNL design required the same propagation delay in each direction. The control of echo is given by the requirement of the echo path delay which is the ROUND TRIP delay of a connection. For echo it doesn't matter if one path has longer dealy than the other only the total delay is important. The delay ind thus loss for different frequencies is of course different and thus a measure called the Weighted Echo Path Loss is generated. WEPL = -20log (1/3200 Integral 200 to 3400 (10**-EPL(f)/20) df From this measure the required loss around the loop may be derived from curves of subjective measurements. This may be partitioned in any manner on transmit and receive paths with no effect on the perception of echo. The VNL plan has been superseded in the digital network by the fixed loss plan since the VNL assumed losses in trunks that do not occur in digital trunks. Rather than place digital pads in the network (and ruin transmission) appropriate analog pads are used at the end points. >While I admit that I have no firsthand experience with >intertoll circuits involving satellites or transoceanic cable, I would >find it difficult to believe that any satisfactory transmission (and >echo) performance could be achieved with the E-W and W-E directions >that have widely *differing* propagation times. Separating the transmit and receive paths between terrestrial and satellite paths lowers the echo path delay (round trip delay) and lowers the requirements on the echo cancellers. Such split path trunks are used. >Obviously, satellite transmission by its very nature imposes >propagation delays which far exceed 45 milliseconds. However, echo >suppressors and intertoll circuit design can be set up to deal with >such increased propagation delays - *provided* that the delay is equal >in each direction. The allocation of delay to tranmit and receive paths is unimportant Only the round trip (echo path) delay is perceptible to the user. >I have never seen anything that split transmission directions >through different facilities. Such facilities do exist.
Utoto@crash.cts.com (Toto uucp) (09/21/90)
In article <12315@accuvax.nwu.edu> ijk@violin.att.com (Ihor J Kinal) writes: >If both sides went over satellite, that would mean a half-second of >extra delay from when one person stopped talking... Would any reader care to confirm or deny the rumor I heard that in the late 1970's AT&T used "satellite avoidance codes" for its inbound WATS (now called "800 Service") customers receiving data calls? As I recall, certain 800-NNX codes were guaranteed a terrestial routing; e.g., 800-223 (NYC Broadway 24) was an avoidance code, but its sister, 800-221, might catch a satellite hop on a transcontinental call. Bill Cerny bill@toto.info.com | attmail: !denwa!bill
harrism@omhftre.raidernet.com (Mark Harris) (09/23/90)
kitty!larry@uunet.uu.net (Larry Lippman) writes: > Unless echo suppressors have become *much* more sophisticated > than those with which I was once familiar, I would be surprised if > such vastly different propagation paths could be used on the E-W and > W-E directions of a given intertoll circuit. I recently took a course in telephony taught by a person with many years experience working for Bell Canada, ATT, and BNR. He stated that, in general, overseas calls try to avoid using a satellite path for both E-W and W-E directions. The reason he gave was to avoid a long delay between, say, a question and a response. Mark Harris UUCP: ...!uunet!mjbtn!raider!omhftre!harrism Domain: harrism@omhftre.raidernet.com