| Summary: | 碩士 === 義守大學 === 電機工程學系 === 88 === Conventional adaptive approaches to signal separation suffer from a number of limitations when applied to multi-user communications. These methods require a reference signal extraction loop or the transmission of training sequence. The blind signal separation have the ability to retrieve the signal, and possibly the channel impulse response, given only the channel output without resorting to any training sequence. There are the cases where the receiver may have no knowledge of the transmitted PN sequence (e.g. when intercepting an unfriendly transmission in low probability intercept (LPI) communications).
Many of the proposed blind techniques usually incorporate constant modulus criterion (CM) which uses the higher order statistical information of receiver signals. But most of them are with low convergence rate and not adequate for real-time implementations.
In this thesis, the multi-rate/multi-channel spread-spectrum model is exploited to derive subspace algorithm based linear blind signal estimation methods. As we reviewed, the subspace algorithm exhibits very good convergence behavior.
Eventually, since consider the existence of time varying channel, adaptive method is exploited to track it efficiently as to the estimation of signals is more successful.
|
|---|
