Dissertation > Excellent graduate degree dissertation topics show

Applications of Cellular Neural Networks in CDMA Communication Systems

Author: ZhaoZuo
Tutor: WuZhiLu
School: Harbin Institute of Technology
Course: Information and Communication Engineering
Keywords: Cellular Neural Network Code Division Multiple Access Multiuser Detection Chaotic Sequence Hyperchaotic Sequence
CLC: TP183
Type: Master's thesis
Year: 2007
Downloads: 105
Quote: 1
Read: Download Dissertation

Abstract

Cellular neural networks(CNNs) are a kind of recurrent neural networks proposed based on Hopfield neural networks and cellular automata, and constitute a class of recurrent and locally coupled arrays of identical dynamical cells, which can be implemented by VLSI easily and be applied to the signal processing. CNNs are also a kind of nonlinear systems, and with proper parameters, the dynamical behavior of a simple CNN system will show interesting bifurcation and complex chaos.Code division multiple access(CDMA) systems become more and more popular in the 3rd generation mobile communication systems due to its attractive features, such as jam-up mitigation, multipath fading mitigation, high security and high spectrum efficiency. The capacity and performance of a CDMA system are essentially dominated by multiple access interference(MAI), which is the result of the nonorthogonality between the spreading sequences of different users. Recently, multiuser detection and better spreading sequences design are the two main methods to mitigate the MAI in CDMA systems.In this dissertation, CNNs have been applied to several methods to mitigate the MAI in CDMA systems, and the hardware realization of CNNs is discussed. First, CNNs have been used in multiuser detection due to their powerful signal process capability. The stochastic CNN(SCNN) detector is proposed here through adding a stochastic term to a CNN, which can avoid local minima in the CNN. The results of simulation show that the SCNN detector has better performance in reducing bit-error rate(BER) and the near-far effect compared with other recurrent neural network multiuser detection techniques. Second, chaotic and hyperchaotic phenomena are generated through the complex dynamical behavior in CNNs. CNN two-value chaotic spreading sequences and multivalue hyperchaotic spreading sequences are designed through sampling and combination. It is shown that these two kinds of CNN chaotic and hyperchaotic sequences are all sensitive to the initial values. Especially, the multivalue hyperchaotic spreading sequences have extremely richer code elements and higher security compared with other conventional chaotic spreading sequences. The CNN two-value and multivalue sequences are filtered out referring to their equilibria points, self-correlation and mutualcorrelation, and then they are applied to CDMA systems. It is shown that the filtered CNN chaotic and hyperchaotic sequences have obvious advantages in reducing BER compared with the conventional spreading sequences such as m sequences and Gold sequences. Last, the hardware realization of CNNs is discussed. A CNN with three cells and its chaotic sequences output are realized based on field programmable gate array(FPGA). The results of simulation show that the realization of the system is feasible, which lays a foundation for the engineering realization of the CNN.

Related Dissertations

CLC: > Industrial Technology > Automation technology,computer technology > Automated basic theory > Artificial intelligence theory > Artificial Neural Networks and Computing
© 2012 www.DissertationTopic.Net Mobile