Asynchronous DS/CDMA systems employing spreading codes generated by chaos maps with Markovity

在DS-CDMA通信中，线性反馈移位寄存器（LFSR）序列，如Gold码和Kasami码，被用作扩频（SS）码。在移动蜂窝系统中，上行信道的传播延迟是随机分布的，同步定时误差无法消除。因此，我们对芯片异步系统进行了评估。我们的研究结果总结如下：1)将总平方相关（CA-TSC）定义为所有用户干扰的总和，并给出其下界，称为Welch界的芯片-异步版。相互干扰取决于扩频码和芯片波形。它们通常是独立设计的，但它们的联合优化应该具有更好的性能。我们放宽了Nyquist条件，该条件保证符号间的干扰为零，通常被认为是满足的，以最小化总干扰。我们发现高斯脉冲和马尔可夫码对比传统的提高余弦脉冲和LFSR序列对具有更好的性能，其中两个脉冲都被调整为具有相同的多余带能量。2)基于后验概率计算，提出了一种新的码采集算法。所提出的方法称为“计数法”，非常简单有效；它只计算接收器输出超过预定阈值的事件数，并且实现比串行搜索更快的代码获取。更重要的是，该方法保证了最大的采集错误率。3)研究了神经网络关联记忆模型与CDMA系统的密切关系。提出了一种适用于芯片异步CDMA系统的联想存储器，该存储器对马尔可夫模式的记忆能力优于id模式

In DS-CDMA communications, linear feedback shift register (LFSR) sequences, such as Gold and Kasami codes, are employed as spread spectrum (SS) codes. In mobile cellar systems, propagation delay is randomly spread for uplink channels and synchronization timing error cannot be eliminated. Hence, we have evaluated the chip-asynchronous systems. Our results are summarized as follows 1) Chip-asynchronous version of total square correlation (CA-TSC) was defined as the total value of all users' interference and its lower bound, named chip-asynchronous version of Welch bound, was presented. Mutual interference depends on spreading codes and chip waveforms. They are normally designed independently but joint optimization of them should give better performance. We relaxed the Nyquist condition, which guarantees zero inter symbol interference and is often assumed to be satisfied, to minimize total interference. We showed that the pair of Gaussian pulse and Markov codes give better performance than the conventional pair of raised cosine pulse and LFSR sequences, where both pulses are adjusted to have the same excess-band energies. 2) Based on a posteriori probability calculations, we proposed a new code acquisition algorithm. The proposed method, called "counting method", is very simple and effective; it counts only the number of events that receiver outputs exceed the pre-determined threshold and it realizes faster code acquisition than the serial search. More importantly, this method guarantees the maximum acquisition error rate. 3) Close relation between the associate memory model of neural networks and CDMA system is investigated. Associative memory corresponding to chip-asynchronous CDMA system was proposed, which was shown to memorize Markov patterns better than i.i.d. patterns