Evolutional Algorithm Based Learning of Time-Varying Multipath Fading Channels for Software Defined Radio

摘要

The software defined radio with reconfigurable signal processing devices requires the learning of multiple unknown parameters to realize the potential capabilities of cognitive communication. Evolutional algorithms are optimal multi dimensional search techniques, and are well known as the effective methods for the parameter learning. Evolutional algorithms can be employed both for learning of the signal parameters and the communication channel parameters. This letter proposes an evolutional algorithm for learning the mobile time-varying channel parameters without any specific assumption of scattering distribution. The proposed method is very simple in the realization, but can provide precise channel estimation results. The simulation results, which are conducted for an example of OFDM communication under the time-varying fading channel, show that the proposed learning method can achieve the better BER performance.