Abstract: In this paper, we proposed a discrete time cellular neural network (DTCNN) equalizer for GSM system and discussed the annealing method for DTCNN, both unannealed and annealed CNN equalizer's performance are simulated on computer. Compared with traditional GSM equalizer based on Viterbi Algorithm, DTCNN equalizer is easily realized. Through hardware annealing, we can set up high quality equalizer based on CNN for GSM system.

Key words: adaptive equalization, GSM system, CNN, hardware annealing

摘要： 提出并讨论了用离散时间细胞神经网络(DTCNN)实现用于GSM系统均衡器的可能性，并基于对均衡算法的分析推出了相应细胞神经网络模板的设计方法，针对存在多径干扰的二值信号处理问题进行了计算机仿真，结果证明了这一想法是合理的。

关键词: 自适应均衡, GSM系统, 细胞神经网络, 硬件退火