lai@vedge.UUCP (David Lai) (10/20/88)
The trailblazer modem slices up the bandwidth and adjusts the usage of the bands depending on the line conditions. I see the TB's main drawback versus v.32 is that thrashes badly in full duplex operation. So how about statistically allocating bandwidth for the send/receive channels depending upon the load? In other words, under an essentially half duplex operation (ie. one way file transfer, or typical user on a keyboard), statistically more data will be sent one way than the other, so allocate more bands to that channel. When the load characteristics are essentially full duplex (ie. two way file transfers), then split the available bandwidth between the two channels. This way one still has the advantages of a fast half duplex modem, while not choking under a full duplex load. Make the algorithm similar to the line adjusting algorithm, continuously adjustable depending on the load of each channel. Sounds do-able, doesn't it? Or am I forgetting something fundamental about signal processing or telephony? Anyone have comments? -- "What is a DJ if he can't scratch?" - Uncle Jamms Army The views expressed are those of the author, and not of Visual Edge, nor Usenet. David Lai (vedge!lai@larry.mcrcim.mcgill.edu || ...watmath!onfcanim!vedge!lai)
dtynan@sultra.UUCP (Der Tynan) (10/22/88)
In article <1653@vedge.UUCP>, lai@vedge.UUCP (David Lai) writes: > > In other words, under an essentially half duplex operation (ie. one way file > transfer, or typical user on a keyboard), statistically more data will be sent > one way than the other, so allocate more bands to that channel. When the > load characteristics are essentially full duplex (ie. two way file transfers), > then split the available bandwidth between the two channels. > > Sounds do-able, doesn't it? Or am I forgetting something fundamental about > signal processing or telephony? Anyone have comments? > -- > David Lai (vedge!lai@larry.mcrcim.mcgill.edu || ...watmath!onfcanim!vedge!lai) This was basically my argument, as well. I think what's important here, is not whether the TB's can do it *now*, but that they could do it in the future, if they wanted. By using a static allocation of bandwidth, as USR does, this is never possible. In fact, I'd go so far as to say that the given bandwidth split that USR has will invariably be the *wrong* one... Using 512 bands like Telebits does is a real sweet idea. Putting the burden of modulation and line-selection in software, is even better. Now we can upgrade things without a soldering iron (:-) If one looks at that biased trash published by USR, one will notice that the demonstration cases were all BBS's. Not a word about 'uucp'-style or 'batch' transfers. - Der -- Reply: dtynan@sultra.UUCP (Der Tynan @ Tynan Computers) {mips,pyramid}!sultra!dtynan Cast a cold eye on life, on death. Horseman, pass by... [WBY]