| Abstract: | We propose an iterative blind interference reduction strategy for short‐burst coded DS‐CDMA systems. The blind strategy works by creating a set of ‘training sequences’ in the receiver that are used as input to an interference reduction algorithm whose task is to produce a corresponding set of equalizers that attempt to recover the desired signal. To maintain a reasonable complexity level we develop a semi‐blind interference reduction algorithm that is capable of equalizing the received signal with a relatively small training sequence length (thus maintaining a small training sequence set). The objective then becomes to determine which equalizer from the generated set gives the best performance (smallest bit error). It is demonstrated that the success of this scheme depends greatly on the ability to find an appropriate criterion for picking the best equalizer. Of the tested criteria, one based on feedback from the decoder (essentially using trellis information) is shown to achieve nearly optimal performance. Copyright © 2005 John Wiley & Sons, Ltd. |
