Abstract:
With the growth of Internet technologies, efficient high-speed data transmission techniques over communication channels have become an important topic for research. The channels used to send the data distort signals in both the amplitude and phase, causing what is known as Intersymbol Interference (ISI). Other factors like thermal noise, impulse noise, cross talk and the nature of the channel itself, cause further distortions to the received symbols. Signal processing techniques used at the receiver, to overcome these interferences, so as to restore the transmitted symbols and recover their information, are referred to as "equalization methods".
The study of non-linear channel equalization in data communications using minimal radial basis function neural network structure, referred to as Minimal Resource Allocation Network (MRAN) are presented here . A parsimonious network is ensured by the MRAN algorithm, which uses online learning and has capability to grow and prune the RBF network's hidden neurons. Compared to earlier methods, the proposed scheme does not have to reduce the channel order first, and fix the model parameters.
This learning algorithm for the network referred to as MRAN, not only allocates a new hidden neuron based on the novelty of observation but also prunes those hidden neuron units which have insignificant contribution to the outputs of the RBF network.
Here in the thesis, performance of MRAN on a number of applications has been tested. These applications consist of problems, which are both static and dynamic systems. The static problems include channel equalization. The problem for dynamic type consists of nonlinear dynamic system identification
