【作者】 王建；

【导师】 张建伟；

【作者基本信息】 大连理工大学 ， 微电子学与固体电子学， 2014， 硕士

【摘要】 三态内容可寻址存储器(TCAM)是在普通二态内容可寻址存储器(BCAM)的基础上发展而来的,它的本质仍然是一种硬件搜索引擎。它具有比普通BCAM更加灵活的储存和搜索功能。正是这种特殊功能的存在使得TCAM的单元结构比普通BCAM的单元结构更加的复杂。现有的针对BCAM的测试算法已经无法满足TCAM的故障检测,现有的针对高速大容量TCAM的故障检测方法在检测覆盖率上和测试复杂度上都有很多不足。然而随着高性能网络路由器对大容量和高速TCAM需求的不断提升,针对高速大容量TCAM的故障测试已成为TCAM的一个研究方向,因此高速大容量TCAM的可测试性设计具有重要的理论意义以及实用价值。本论文对TCAM的故障检测问题进行了初步的研究,并提出了两种针对高速大容量TCAM的故障检测算法和检测电路的设计。首先,基于现有的公认度比较好的March-Like算法提出了改进型算法。该改进型算法利用匹配信号Hit和优先地址编码器只需2N次写操作、2B+4次搜索操作就可以完成对N*B比特TCAM存储单元的故障检测。在故障检测覆盖率上同March-Like算法是一样的,都可以完成TCAM单元内故障和延迟故障的100%全覆盖检测。但是在测试复杂度上相比于以前的March-Like算法降低了60%以上。再者,从提高单元间故障检测覆盖率方面,本文又提出了一种全新的针对高速大容量TCAM的QZDTest故障检测方法。该方法在故障检测覆盖率方面不仅可以完成TCAM单元内故障和延迟故障的100%全覆盖检测,还可以完成TCAM单元间故障的100%全覆盖检测；在检测速度方面,该方法使用自己的测试向量发生器产生测试向量,通过5N次写操作,6B+12次搜索操作可以对N*B比特TCAM存储单元进行类似于实际应用的快速连续测试。另外,在检测电路的设计方面,通过有限状态机控制移位寄存器来完成对TCAM存储单元的储存和比较操作,充分降低了测试电路的开销。在判定电路设计方面,通过对SA(敏感放大器)和匹配信号发生器之间插入一级选择取反电路,可以将Hit信号作为TCAM存在故障的判定信号,并通过优先地址编码器直接输出优先级最高的故障地址,因此可以完成TCAM的全自动测试。通过原理图仿真验证结果表明该算法确实可以在单时钟周期完成TCAM存储单元的全覆盖检测。本文提出的TCAM可测试性设计非常适用于大容量高速的TCAM存储单元故障检测,如高性能网络路由器中的报文转发、地址分类、网络防火墙以及虚拟网络过滤等专用高级网络应用方面的TCAM故障检测。更多还原

【Abstract】 Ternary content addressable memory (TCAM) is essentially a hardware-based search engine; and it was presented on the basis of Binary content addressable memory. It has more flexible than ordinary BCAM in storage and search capabilities. The existence of such a special function of TCAM makes it more complex than the structure of BCAM. The existed test algorithms for BCAM error detection has been unable to meet the demand of TCAM, the existed error detection method for high-speed large-capacity TCAM have a lot of shortcomings on the test complexity and detection coverage. However, with the demand of high-performance network router for large-capacity and high-speed TCAM continuous improve error detection for high-speed and large-capacity TCAM has become a key research direction, so the proposed high-speed and large-capacity TCAM testability design has important theoretical and practical value.In this paper, a preliminary study of error detection problem of TCAM is carried out, and two error detection algorithms and a detection circuitry are proposed. First of all, a new improved algorithm based on existing March-Like is proposed. The proposed algorithm requires2N write operations and2B+4compare operations to cover100%of targeted comparison faults and delay faults for an N*B-bit TCAM via Hit and Priority Address Encoder outputs. The proposed algorithm reduced over60%compared to the previous March-Like algorithms. Moreover, in terms of improve error detection coverage; the paper put forward a new QZDTest error detection algorithm for high-speed large-capacity TCAM. The new kind of method in the aspect of error detection can not only cover100%inner-cell TCAM cell faults and delay faults but also can cover100%intercell TCAM faults. In the term of test speed:the proposed algorithm use their own test vector generator te^t vectors to complete N*B-bit TCAM error detection through5N write operation and6B+12search operation, the test speed is the same as practical application. Further, in the design of the detection circuit, it use finite state machine to control the shift register to complete the TCAM storage and comparison operations, substantially reduces the cost of the test circuit. In the design of judge circuit, by inserting an inverter and a selection circuit after amplifier, it can get error signal through Hit determination signal and output the highest priority error address directly through the address priority encoder. Circuit simulation results show that the algorithm can fully coverage TCAM detection in a single cycle.The proposed design for testability is very suitable for error detection in large-capacity and high-speed TCAM storage unit. Such as high-performance network router, packet forwarding, address classification, network firewalls, virtual private network filtering and other advanced network applications in TCAM error detection.更多还原