Super fast and efficient channel equalizer architecture based on neural network

Broadband wireless communication systems are currently in a rapid evolutionary phase in terms of development of various technologies, development of various applications, deployment of various services and generation of many important standards in the field. Ever increasing demand on various services justifies the need for the transmission of data at the highest possible data rates. The multipath and fading characteristics of the wireless channels result in various impairments and distortions, the most important of those being the Inter-Symbol Interference (ISI) especially at relatively high data rates. Among the various possible solutions to mitigate ISI, the adaptive equalizer remains one of the most attractive solutions, particularly the algorithms requiring minimal or no training sequence and at the same time are computationally efficient. This paper presents a novel neural networks based architecture for channel equalizers that require only order of 20-40 training symbols to converge to the optimum solution and at the same time is computationally efficient.