Font Size: a A A

Research On Test Scheduling For SOC Based On Multiple Voltage Island

Posted on:2017-01-12Degree:MasterType:Thesis
Country:ChinaCandidate:X L BianFull Text:PDF
GTID:2308330509457059Subject:Instrumentation engineering
Abstract/Summary:PDF Full Text Request
With the increasing integration and the functional complexity of SOC(System on Chip), low power design techniques have become mainstream SOC design concept. SOC based on MVI(Multiple Voltage Island) uses DVS(Dynamic Voltage Scaling) technology to minimize the system energy consumption. However, the testing challenges of SOC based on MVI presented have severely baffled its development. On the premise of high fault coverage, efficient testing of SOC based on MVI in system level is an academic circles’ s research hot topic and the main problem.The testing of SOC based on MVI has the characteristics of the heavy multitaskers and vast constraints between test tasks. The study aimed to improve testing efficiency of SOC based on MVI is developed from the following aspects: the test time optimization of the smallest unit, the test resources allocation optimization and related constraints between different tasks.The reusable IP core is the smallest unit of SOC based on MVI. To achieve the test application time reduced effectively, this paper researches the balance design of scan chains for IP core and proposes a new twice-assigned method based on dynamic adjustment scan chain and mean scan chain. The experimental results on ITC’02 standard test data show that the approach proposed in the paper can effectively reduce the test time of the smallest unit in SOC based on MVI.It is an effective way to improve testing efficiency that the testing resources optimize as well. A good resource scheduling strategy can greatly improve the performance in testing parallel by proper distributions of the limited test resources. DE(Differential Evolution Algorithm) presented in the paper is used for resolving the test resources optimal problem. The probability operator and hybrid mutation strategy can guarantee the high efficiency of DE algorithm. The comparative experiment results verify DE efficient property for the test resources allocation optimization.Test scheduling of SOC based on MVI needs to consider the constrained condition between different tasks based on the two aspects mentioned above and then puts test position and test order for all tasks. The heuristic approach and iterative DE algorithm presented in the paper are used to solve the problem of the related constraints between different tasks respectively. The comparison results show that iterative DE algorithm is superior to the heuristic approach.Test scheduling strategy of SOC based on MVI can effectively reduce the test time in system level without any additional hardware overhead.
Keywords/Search Tags:SOC based on MVI, the balance design of scan chains, test scheduling strategy, differential evolution algorithm
PDF Full Text Request
Related items