Font Size: a A A

Research On Complementary-alternating Self-recovering Controllers

Posted on:2011-03-26Degree:MasterType:Thesis
Country:ChinaCandidate:X M ChenFull Text:PDF
GTID:2178360308973181Subject:Computer system architecture
Abstract/Summary:PDF Full Text Request
As the feature size of integrated circuits (ICs) decreases to nm, the working voltage is dropping constantly, the integrated density and the working frequency is increasing dramatically. Consequently, the soft error rate (SER) becomes higher and higher. Ionizing particles induce SEUs in controllers. As one of the critical applications of the ICs, controllers play significant roles in micro processors, whose capacity of soft error tolerance directly determines the robustness of micro processors, even the whole system. Logic becomes the largest contributor to SER. Protecting logic is not as simple as protecting memories. The cost in area, speed, and power can be significant. Reduction of this cost is therefore mandatory in order to make it acceptable of ICs destined to commercial applications.There are two types of soft errors, SEU (Single Event Upset) and SET (Single Event Transient). The research of SEU tolerant controllers design is developed in the dissertation, the main work of which is showed as follows.(1) A self-recovering method based on dual modular redundancy is proposed in this dissertation. The two redundancies work in a parallel way. When an error occurs in one of the two redundancies, a hardware rollback operation will be automatically performed using the correct state in the other redundancy module. And we add C-element after the output of the two redundancies, thus it can prevent the error from being output. Experimental results show that, this method can provide 99.32% protection from SEU. And, compared to published self-recovering methods based on dual modular redundancy, the area overhead of the proposed method is low, while the increase of delay is negligible.(2) A fast fault-tolerant method is proposed in this dissertation, by which the original FSM is decomposed into two sub-FSMs. The two sub-FSMs work by turns. When an error occurs in one of the two sub-FSMs, a hardware rollback operation will be automatically performed using the correct state in another sub-FSM. Experimental results show that, this method can provide 99.32% protection from SEU, while achieving 13.16% delay reduction on the average.
Keywords/Search Tags:soft error, dual modular redundancy, finite state machine, self-recovering, alternating-complementary
PDF Full Text Request
Related items