Research On Theory And Algorithm Of Blind Equalization Based On High-order Statistics

Nowadays, the world has already entered the information age developing at full speed, communication has become the front discipline which is a trade with the fasted developing speed in an information industry. In digital communication systems, intersymbol interference (ISI) which is an important factor that degrades the performance of communication is encountered inevitably when unknown signals are transmitted through non-ideal channels. In order to reduce ISI, equalize the characteristic of channels and recover transmitted signals, equalizer have to be used in the receiver. Periodic training sequence have to be transmitted repeatedly for the traditional adaptive equalizers which wastes plenty of bandwidth , decreases the transmission rate and increases the system complexity. Blind equalization is a new adaptive technique without resorting to a training sequence, which only relies on the prior knowledge of transmitted signals to equalize the channel characteristic, and makes sure that the output sequence can approximates the transmitted signals as accurate as possible. Therefore , it is more efficient and more applicable to advanced communication systems. Recently, Blind equalization which is very important for instructing both the theory and the applications in communication, radar, sonar , control engineering, earthquake prospecting, biomedicine engineering and so on, has been becoming a very popular research topic in communication system.The dissertation focuses on the performance of blind equalization algorithm using high-order statistics(HOS) implicitly and explicitly. Blind equalization algorithm based on HOS was presented and developed in 80's. HOS convey not only amplitude but also complete phase information, so relation equation between HOS of the signal and channel parameters can be established only using output signal, then obtaining the parameters in a manner to solve the equations. Meanwhile, because the high-order cumulant of Gaussian noise is zero, the non-Gaussian signals can be extracted from Gaussian noise using HOS, moreover, high-order cumulant can detect and deal with non-linear problem effectively. However, this algorithm has high...