Statistical spectral analysis: a nonprobabilistic theory
Statistical spectral analysis: a nonprobabilistic theory
Exploitation of cyclostationarity for identifying the Volterra kernels of nonlinear systems
IEEE Transactions on Information Theory
| Hi-index | 0.00 |
A new set of efficient blind nonlinear channel identification methods is presented that exploits the cyclostationary nature of signals in many applications. The methods are of the batch type, use only the cyclic autocorrelation or cyclic cepstrum of the received signal, and can be used for real or complex (causal or noncausal) finite-memory quadratic Volterra models. In most cases, the solutions are direct and no equations need to be solved. Identification of various nonlinear models is discussed with the fractionally spaced impulse data train input. A simple method is considered for estimating the memory in a quadratic Volterra model. It is shown that the proposed methods can identify the nonminimum phase linear subsystems of the nonlinear channel model. Computer simulation results support the theory.