Decoding a signal for 2 users simultaneously in CDMA?

[coolguy_ar]

Hi,
This is actually my major project and I'm having problems with being able to decode the incoming signal of 2 or more different users, "at the same time". If their signals are all added up, then how is the correlation with PN sequence going to help!

We have basically designed a spatially intelligent antenna array, and get the DOA of the signals coming in...

One more clarification. How many signals can I take in at one in an array with no. of elements = 4(say), and be able to use Music Algo on it to be able to detect the direction of arrival?!

Please help me out asap.

 

[iTdl]

(just an idea)
As the CDMA codes are orthogonal. If we correlate all the PN codes with received signal (considering that received signal is time aligned. Otherwise algo needs extension to consider delay or non-alignment), we will get different power level for all the available PN codes.

Now set a threshold, from knowledge base, to consider valid signals.
Consider case of two signals only. Then you should be able to decode the signals, in principle.

Just try simple Matlab code for this.
I will also try on weekend.

 

[coolguy_ar]

Actually, I've got to model a fading channel, and then pass either one signal, or if possible two signals, and receive them at an array of antennas.

I need a Matlab program to first generate the pn sequences for different users. But the problem lies in that, how each code upon arrival in the form of a total sum of signals is extracted. And then, I'll need to model using a Rician and a Rayleigh channel separately,which I'm working on at the moment.

I also have another doubt if the fading channel is to be added before the antenna detects the DOA or even before the angle is estimated. I can provid you with my code regarding the exact problems, if you want.

 

[iTdl]

1. you dont need to extract the signal at the receiver ( i mean no physical operation required for extraction before correlation)
i) when you do the correlation of available PN sequence codes from set (user is for sure going to send code from this set) with the received signal (which is sum of all orthogonal PN seq codes). Store power after each correlation. you will find the correlation (i.e.power) higher than threshold. The codes having higher power are supposed to be transmitted.
ii) Question remains who transmitted what. That is further solved N/W layer. I am not very confident in that.

2. About fading channels, Simulink models are readily available. you can try those. Matlab help/manual should be sufficient for this.

 

[coolguy_ar]

Hi,
I am only required now to implement the fading channel. I'm not going to go for the simulink, since I have a direct function "ricianchan" or "rayleighchan".

But I have a question... How does this really affect the performance of the antenna array?? Does it really affect it? The signal only gets some changes in power n phase... How does this reduce the likelihood of being identified by an antenna array? Especially when I am using algorithms like MUSIC, or ESPRIT...

There's one more question... Is there an upper limit on the no. of signals that can be resolved using MUSIC algorithm? for eg. how many incoming signals's DOA can be estimated with an array of say 4 or 9 elements using the MUSIC algorithm?

Added after 10 minutes:

Hey, there's one more help I need.. I need to model a channel with 2 or 3 multiple paths that are rayleigh fading in nature. What are the best values for the delay and the power profiles to be used in the above functions??