Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : George Gilder - Forbes ASAP -- Ignore unavailable to you. Want to Upgrade?

To: D.J.Smyth who wrote (1144)	3/31/1999 10:48:00 AM
From: George Gilder	Read Replies (3) \| Respond to of 5853

Thank you very much for your pointers on correlation and matched filters. However, the two urls are the same.

To: D.J.Smyth who wrote (1144)	3/31/1999 10:52:00 AM
From: Clarksterh	Read Replies (1) \| Respond to of 5853

Darrell - It's truly not as simple as using different spark plugs. You've taken the analogy outside of its design specs. I was only referring to the fact that matched filters are used everywhere (there is probably at least one in your modem and CDMAOne probably uses several in any of their chips). To talk about matched filters without the architecture is almost meaningless and anybody who claims to be technically knowledgable and does so loses some credibility. Claims that a matched filter in-and-of-itself will solve the near/far problem are, for the most part, huey. However, there are technologies which will increase the capacity of unsynchronized CDMA and may minimize some of the near/far problems and your URL points to the most important of these technologies - MUD. (which is almost completely separate from the topic of matched filters) I discussed MUD several months ago on the Qualcomm thread, but, given the huge volume there, it might be appropriate to recap. Normally in unsynchronised CDMA (the reverse link only in CDMAOne) the various different users interfere with each other - User A contributes significantly to User B's noise and vise versa. The most important parameter in any communication system is the signal to noise ratio, so if you can reduce the noise that each user is seeing you can add more users and worry less about somebody else's signal drowning out yours (i.e. Near/far). With MUD the system detects each users signal and actually subtracts it out of the noise for the other users. To my knowledge this is not used in commercial systems yet because it is a huge processing hog. But as processing power increases this will become easier to do. However I know of no particular reason why Qualcomm couldn't implement the same thing and in fact I'd be surprised if they didn't sometime in the next year or two. Clark PS George, FYI a follow up on the limits of MUD: In order for MUD to work the processor needs to know sort-of what all the signals look like and the signals have to be 'separable'. If you have spreading of 8 times, meaning that each data bit has 8 chips in it, then there must be significantly less than 2^8 separable signals possible (I'd have to review/learn some more theory to figure out exactly how many). This places a limit on the number of bits/MHz even with MUD. Just FYI. PPS Darrell, note that your definition of a matched filter (a matched filter ... doesn't know when a signal is arriving (is activated when a signal or wave arrives)) is a reasonable one, but what reasons does he give for Qualcomm not being able to retrofit? Also note that in this regard the primary benefit of a matched filter is a quicker lock up. This is good, especially in a packetized system with lots of unpredictable voids, but it isn't clear how it provides much beyond that.