ITR/SI Collaborative Research: Investigations on CDMA Systems using Multiple Level Sequences and Partitioned Reduced Complexity Multiuser Detection Receivers

自1994年在西雅图部署第一个IS-95系统以来，码分多址系统在无线通信领域取得了巨大的商业成功。事实上，大多数未来的无线通信系统都将使用码分多址技术。然而，为了充分发挥码分多址在更高的系统容量、更低的误码率和更低的丢包率方面的巨大潜力，在多址干扰下的解调问题仍然是所有码分多址系统的重要瓶颈。为了解决这个多址干扰问题，研究人员和工程师们自20世纪80年代初S提出了所谓的多用户检测器(MUD)。不幸的是，到目前为止，由于这些结构的令人望而却步的复杂性，还没有在实际的CDMA系统中实现这些多用户检测器。造成这种情况的主要因素有两个。首先，由于几个实际原因，码分多址系统必须使用长扩频序列，而多用户检测器必须使用短扩频序列。其次，即使是最简单的多用户检测器的复杂性也太大，无法在最快的电子设备中实现，除了最小的数据速率。在这项工作中，我们建议通过合作努力解决这两个问题。合作的安排将使得密歇根州罗切斯特市奥克兰大学的刘庆冲博士最初将专注于第一个目标，而俄亥俄州雅典俄亥俄大学的大卫·W·马托拉克博士将最初(同时)专注于第二个目标。结果将经常交流，会议将每季度举行一次。经过一些早期的进展，随着我们结合工作目标和考虑到系统(发射器和接收器)的性能，我们的合作将变得更加紧密。我们首先推广了从短序列构造长扩频序列的一些最新结果。这可以称为系统目标。这些新序列既适用于传统的码分多址系统，也适用于多用户检测器。这种方法是由我们中的一个人发明的，并已由休斯电子公司在Spaceway系统中运行速度可能为400 Mbps的最快无线网络中实现。它显著降低了接收机的复杂度和成本，并提供了基本最优的性能。通过在多个层次上测量序列相关性，该方法为序列设计提供了新的见解，极大地降低了宽带无线通信信号设计和检测的复杂性。这种方法有可能帮助弥合当前的码分多址系统和多用户检测器之间的差距。我们还建议从降低复杂度的网格搜索技术的角度，结合最有前景的MUD接收机结构之一，以分区的方式来研究降低复杂度的多用户接收机。这可以被称为接收者目标。MUD接收机是最小均方误差(MMSE)接收机，由于其良好的性能和适中的复杂度而备受关注。分区方法的目的是在MMS前端和低复杂度的网格处理器之间分担检测任务。这些接收机将使用未来第三代和第四代码分多址无线通信系统标准中指定的新闻阅读序列，以及在系统目标中开发的多级序列。降低复杂度的网格搜索技术将探索将已研究的类似技术用于重构化，但不完全应用于码分多址多用户检测问题。我们这两个目标的目标是为现有的和计划中的码分多址系统与理论上的多用户检测器建立一座基本的桥梁，从而克服码分多址系统中多址干扰问题的瓶颈。这项拟议工作的两个目标都将使用分析，然后是计算机模拟。这两个目标也将需要研究生和本科生的帮助。建模工作的目的是使理论与实际实施相协调，从而提供最佳意义上的工程教育：将原理与实践联系起来。培训学生在码分多址的系统设计和接收器设计方面也自然会对无线产业有价值。此外，这项研究将为两所大学的几门新的研究生课程和本科生设计项目提供材料。

CDMA systems have achieved great commercial success in wireless communications since the firstIS-95 system was deployed in Seattle in 1994. In fact, most of the future wireless communicationsystems will be using CDMA technology. Yet, as far as reaching the great potential of CDMA in termsof higher system capacity, lower bit error rate, and lower packet loss rate, the problem of demodulation inthe presence of multiple access interference remains as a significant bottleneck for all CDMA systems.To solve this multiple access interference problem, researchers and engineers have worked since the early1980's on so-called multiuser detectors (MUD). Unfortunately, thus far none of the multiuser detectorshas been implemented in a real CDMA system, because of the prohibitive complexity of these structures. There are two major factors that contribute to this situation. First, CDMA systems must use longspreading sequences for several practical reasons, whereas multiuser detectors must use short spreadingsequences. Second, the complexity of even the simplest multiuser detector is still too great to beimplemented in the fastest electronics, for all but the smallest of data rates. In this work, we propose toaddress both of these problems in a collaborative effort. The collaboration will be organized so that Dr.Qingchong Liu of Oakland University in Rochester, Michigan, will initially focus on the first objective,and Dr. David W. Matolak of Ohio University in Athens, Ohio, will initially (and concurrently) focus onthe second objective. Results will be exchanged frequently, and meetings will be held quarterly. Aftersome early progress, our collaboration will grow closer as we combine the work objectives and considersystem (transmitter and receiver) performance. We propose first to extend some recent results obtained for the construction of long spreadingsequences from short sequences. This can be termed the system objective. These new sequences willbe appropriate for both conventional CDMA systems and for multiuser detectors. This method wasinvented by one of us and has been implemented in possibly the fastest wireless network running at400Mbps in the Spaceway system by Hughes Electronics. It has significantly reduced receivercomplexity and cost, and gives essentially optimum performance. By measuring sequence correlations atmultiple levels, the method provides new insights on sequence design and tremendously reducescomplexity in signal design and detection for broadband wireless communications. This method has thepotential to help bridge a gap between current CDMA systems and multiuser detectors. We also propose to study reduced complexity multiuser receivers from the perspective of reducedcomplexity trellis search techniques, combined with one of the most promising MUD receiver structures,in a partitioned manner. This can be termed the receiver objective. The MUD receiver of interest isthe minimum mean-squared error (MMSE) receiver, which is attractive for its good performance andmodest complexity. The partitioning approach will aim to share the detection tasks between the MMSEfront-end and the reduced-complexity trellis processor. These receivers will make use of the newspreading sequences specified for use in future 3 rd and 4 th -generation CDMA wireless communicationsystem standards, and the multiple level sequences developed in the system objective. The reducedcomplexity trellis search techniques will explore use of the analogous techniques researched forequalization, but not fully applied to the problem of CDMA multiuser detection. Our goal for these twoobjectives is to create a fundamental bridge to connect the existing and planned CDMA systems with thetheoretical multiuser detectors so that the bottleneck of the multiple access interference problem inCDMA systems can be surmounted. Both objectives of this proposed work will employ analysis, followed by computer simulations.Both objectives will also require the assistance of graduate and undergraduate students. The modelingwork will aim to reconcile theory with practical implementation and thus provide engineering educationin the best sense: connecting principles and practice. Training students in both system design andreceiver design for CDMA will also naturally be valuable to the wireless industry. In addition, theresearch will provide material for inclusion in several new graduate courses, and in undergraduate designprojects, at both universities.