Code Division Multiple Access (CDMA) continued [Under Multiple Access]
The spreading codes that we have seen till now for DS-CDMA are considered (mutually) orthogonal due to their specific characteristic of zero correlation. Popular examples of CDMA codes are 64 chips long Walsh codes, used in IS-95/cdmaOne and cdma2000. Haddmard matrices are used to generate the Walsh codes.
Consider following example of two Walsh codes (4 chip long) - in general N chip code Walsh code will support N multiple users.
W(1,4) = 1 0 1 0
W(2,4) = 1 1 0 0
If we Ex-OR, we get 0 1 1 0, Two 0s and two 1s, and so these codes are indeed Orthogonal.
Note that for these orthogonal codes to work, transmitter and receiver need to be in sync. And so this method is also known as Synchronous CDMA.
But in practice like in mobile handset scneario, we may not have this perfect synchronisation. Here instead of Walsh codes, "pseudo-random" or "pseuo-noise" (PN) codes are used. Long code and Short code used in IS-95/cdmaOne and cdma2000 are PN codes.
PN codes are seemingly random, but can be generated in an algorithmic way. These codes, though not orthogonal, have low correlation with each other and its own time shifted version. In addition, large number of PN codes resemble ("Gaussian") noise. So it is possible for receiver to extract the intended signal by iterating through possible PN codes.
This method is known as Asynchronous CDMA and used in commercial wireless CDMA systems.
This completes brief on CDMA. 3GPP has adopted W-CDMA (called Wide due higher chip rate of 3.84 Mcps).
Further reading would be 1) generation of PN codes 2) how short/long codes are used in commercial CDMA systems like IS-95/cdmaOne, cdma2000 etc. 3) Power control etc.
References: Intro to CDMA from Scott Baxter, CDMA tutorial from complex2real site, Wiki page on CDMA, and Wiki page on Walsh codes.
© Copyright Samir Amberkar 2010| Multiple Access (CDMA - 2) « | Multiple Access Index | . |