Re: Noise effects overall data transmission speed...nothing more nothing less.
Matt,
I'm very busy at the Cape, but there are at least a few issues
you need help with.
If I can run at T1 speed [CAP] vs 7Mbps [DMT] who cares?
Let's get you straight, both CAP and DMT can run at 7Mbps, both are rate adaptive and can run anywhere between 1.5 and 7Mbs, depending on Loop Length. DMT can run at 8Mbps even 10Mbps on a short loops.
Both have a lite version that can run at 1.5 Mbps.
The Lite versions being splitterless, again this is a plus for DMT, and worse for CAP.
You seem to think noise does not matter, it does and I'll get to your Packet Retransmission issue in a minute.
In the Splitterless (Lite) world your data travels though all the Twisted Pairs though out your Home, and you pick up all kinds of noise. See previous Links. If you got a CAP modem you'll need lots of those retransmission of packets.
The TCP/IP retransmission count on Win NT/95 has a default value of 5. IMHO - Plan on increasing the default if your using Splitterless CAP. Retransmission does eventually time out and you may miss the scene where the Hero of the movie kills the Bad Guy.
What comes with more bandwidth? Larger files like real-time video or VOD, and not the 30 second videos we play today. Packets break up the files, so retransmission may be less time consuming than retransmission of the entire file, but you still need to get the entire large file in a reasonable time otherwise there will not be a smooth video seen. They buffer audio today so it sounds smooth, lot's of retransmission going on, and we still get gaps. With video you would need GigaBytes of buffer memory for video.
Let me quote pieces of an article as a reference to the current limitations of TCP/IP on just simple audio.
A major drawback of TCP in dealing with audio streams, however,
is that if a packet is lost, the server will need to retransmit it. And because the packet is significantly larger than a UDP packet, it
also takes longer to retransmit. Therefore, while a lost UDP packet
results in a tiny gap in the audio transmission, a lost TCP packet
knocks a huge hole, probably causing audio playback to stop until
the packet is retransmitted and successfully delivered. This packet
retransmission also results in higher overhead on the server than
UDP-based solutions.
In general, UDP is best used in momentary audio-delivery
applications such as Internet phone products. TCP strikes a happy
medium, perhaps the best all-purpose solution despite its inherent
limitations due to packet size and retransmission factors. IP
multicasting is ideal in situations where the same audio stream will
be delivered to many people simultaneously. While there's no clear
winner at this time, we have good solutions to meet specific needs.
The assumption is that the buffer will make up for transmission delays, which could result in audio gaps until new data is delivered to the player.
zdnet.com
Noise in ADSL transmission is a large factor, remember we are not dealing with Laboratory conditions in the real world, the greater percentage of POT's lines are 20, 30 years old or more in the U.S., and older in many places in the world. DMT was designed to handle these conditions.
There are other conditions that effect POT'S lines as well, again DMT was designed to handle these as well. Example:
"In theory, CAP, which is essentially equivalent to QAM, and DMT will have exactly the same performance when you are allowed to have INFINITE complexity.
In practice, of course, you can't have infinite complexity, so which technique is better depends on the transmission environment. In the case of ADSL, DMT is far the superior technology, because the copper line is slowly time varying (due to temperature variation, crosstalk variation, etc.), contains frequency notches due to bridged taps, and suffers from impulse noise hits. All these impairments are difficult for CAP or QAM to deal with using a reasonable length equalizer, whether it's linear or a DFE."
For many of the factors explained above and in my previous links, the ETSI and ANSI choose DMT for the standard.
In order for CAP, even to be used as an ADSL transmission medium, they had to force the establishment of an ADHOC committee under the ANSI T1E1.4
CAP would not even be in consideration if not for the early backing with lot's of $, and lies in CAP's defense, from the company that developed it, I'm sure someone will tell you who that CO. is if you don't know, and want to even bother asking.
I'm not here to bash Westell, at one time I thought they had a Great Chance, especially when they had a deal going to license DMT at one time. I have not keep up with them. Aware even though they are small, always had ADI on their side, and now at least have some Intellectual Property in G.Lite. PairGain in my view with it's current price is a good value, and Orckit seems to have a very firm grasp as well.
As the ADSL (Lite) Hype hits early next year you could see another rally in ADSL stocks.
I was in on the first round. #'s approximate and go from Feb. 96 to June 96.
In 5 months one could have made lot's of $$.
PAIRGAIN went from $20. to $120 then Split 2 for 1
Westell went from $12. to $80 then split 2 for 1
Amati went from $6. to $36.
Good luck to all who own ADSL stocks during the next rally.
Oh, Matt, I'm not finished yet!
The long and the short of all this technical gobbldygook is still speed. No one is doubting that DMT is faster....it is much faster. The DSP technology is remarkable and these folks are brilliant for creating such great technology.
1) Again both CAP and DMT run Fast.
2) Westell did not create the technology. They use CAP chips which are from that other company I mentioned.
As I stated, someone will tell you if you don't know the name, they lied so much on paper and to my face at SuperComm and N+I , I won't say the name. When I played my cards, after letting them Ramble on about CAP for 20 minutes, you should have seen these 2 guys. Then one says " Are you with Amati?" No I said just a well informed investor, as I turned on my heels and left.
Keep up the DD and you'll get there Matt, kudos for finding and investing in ADSL, in the first place. The Stocks will rally again, and you'll be on the Train.
Westell may do OK in the future, but I have not followed them enough to give anyone advice.
JW@KSC |
