Dynamic Spectrum Management Project



Multiuser Detection

Most multiuser communications systems like DSL and wireless are essentially limited by interference (crosstalk) between users. So, the multiuser detector is often an interference canceller. The theoretically optimal solution for interference cancellation, as for any other detection problem, involves maximizing the a posteriori probability of the primary signal (MAP rule). But, this often results in a receiver too complex for practical use. Our group has been working on sub optimal receivers that can provide significant performance enhancement without excessive complexity.

In a DSL system, the crosstalk can be classified into near-end crosstalk (NEXT) and far-end crosstalk (FEXT). Crosstalk cancellation for DSL systems that are constrained by FEXT were implemented as part of Vectored Transmission schemes. However, NEXT from other services remains one of the major impairments in current ADSL downstream transmission. This crosstalk signal has large excess bandwidth, and its spectra in the main lobe and the excess band are strongly correlated. This correlation provides the opportunity to cancel the crosstalk signal in some dependent frequency bands. For example, crosstalk can be estimated in the excess band and canceled in the main lobe or vice-versa. Application of this technique can lead to a much simplified cross talk cancellation scheme. Reference [1] describes two such methods. It also describes a fast computational algorithm for practical implementation with slightly degraded performance.

Reference [2] describes a soft linear canceler to reduce interference on a transmission system by another with an overlapping frequency band. This low complexity method performs almost as well as a ML detector, with a small loss of bandwidth in the system of interest. The method is demonstrated on a VDSL system with home LAN intereference.

Similar methods can be developed for cancelling Co-Channel Interference (CCI) in a wireless multiuser system. References [3], [4] describe a sub-optimal multiuser detection scheme that outperforms beamforming and space-time equalization. More on this and other methods for wireless are given in the Wireless Research section.
 


http://cafe.stanford.edu/~cioffi/dsm/