Multiuser detection and interference suppression techniques for direct sequence spread spectrum systems in which all users employ same spreading code

1. A method

for separating signals from multiple users in a direct sequence spread spectrum communication system in which said multiple users each use a substantially equal spreading code, said method comprising steps of, at a receiver coupled to said communication system:

receiving a signal modulated by symbols;

partitioning the signal into a set of delays;

determining a correlation statistic between the delays;

comparing the correlation statistic to a predetermined value wherein a high correlation indicates the delays are from the same user;

identifying a plurality of significant signal components in response to successive steps of interference suppression; and

associating, each with a distinct user, one or more of said significant signal components;

wherein

said steps of successive interference suppression include steps of

identifying a delay of signal components having a largest correlation;

adding the delay of the signal component having the largest correlation to a current active set of delays, active set of delays being a set of delays of the significant signal components;

updating the current active subspace by adding spreading sequence shifted by the delay of the signal component having largest correlation; and

projecting the received vector orthogonal to current active subspace.

2. A method as in claim 1 wherein said step of determining a correlation statistic includes the steps of:

generating a sequence of demodulation statistics, each said demodulation statistic corresponding to a delay in an active set of delays;

computing a demodulation statistic vector corresponding to said sequence of demodulation statistics;

computing correlation statistic between all demodulation statistic vectors considering a pair of delays in the active set of delays at a time.

3. The method of claim 1 further including:

applying said delays having a correlation statistic greater than or equal said predetermined value to a rake receiver associated with said same user.

4. A method as in claim 1 , wherein

said steps of successive interference suppression are performed during a training signal;

wherein the training signal consists of training symbols modulating the spreading code, each training symbol consisting of chips.

5. A method as in claim 1 , wherein

said steps of identifying significant signal components include steps of

associating multi-path components in an active set of delays that are with a predetermined number of chips apart from another user.

6. A method

for separating signals from multiple users in a direct sequence spread spectrum communication system in which said multiple users each use a substantially equal spreading code, said method comprising steps of, at a receiver coupled to said communication system:

receiving a signal modulated by symbols;

partitioning the signal into a set of delays;

determining a correlation statistic between the delays;

comparing the correlation statistic to a predetermined value wherein a high correlation indicates the delays are from the same user;

identifying a plurality of significant signal components in response to successive steps of interference suppression; and

associating, each with a distinct user, one or more of said significant signal components;

wherein

said steps of identifying significant signal components include steps of

identifying one or more correlations within the signal, and

associating the identified correlations with distinct users, each said distinct user being identified by one or more separate multi-path components;

and wherein

said steps of identifying one or more correlations include steps of

computing inner product of the linear minimum mean squared error correlation with spreading sequence separated by d chips, where d ranges over all possible chip offsets that may correspond to a complete symbol falling into observation interval;

identifying delay for which output of the inner product is maximized;

identifying all delays, as part of an active set of delays, for which output of the inner product is more than a predetermined threshold;

associating the delays that are within a predetermined number of chips apart from the delay for which the output of the inner product is maximized with a primary user;

estimating interference subspace spanned by the delays that are farther than the predetermined number of chips apart from the delay for which the output of the inner product is maximized, the interference subspace comprising of contributions from secondary users; and

recomputing a correlation locked to the primary user that minimized mean squared error, subject to the constraint that the correlation is orthogonal to the interference subspace.

7. A method as in claim 6 , wherein said step of determining a correlation statistic includes the steps of:

generating a sequence of demodulation statistics, each said demodulation statistic corresponding to a delay in an active set of delays;

computing a demodulation statistic vector corresponding to said sequence of demodulation statistics;

computing correlation statistic between all demodulation statistic vectors considering a pair of delays in the active set of delays at a time.

8. A method as in claim 6 , further including:

applying said delays having a correlation statistic greater than or equal said predetermined value to a rake receiver associated with said same user.

9. A method as in claim 6 , wherein

said steps of successive interference suppression are performed during a training signal;

wherein the training signal consists of training symbols modulating the spreading code, each training symbol consisting of chips.

10. A method as in claim 6 , wherein

said steps of identifying significant signal components include steps of

associating multi-path components in an active set of delays that are with a predetermined number of chips apart from another user.