A gigantic amount of test pattern covering all faults is required to attain a high test quality in modern VLSI testing. With this, the Test Data Volume (TDV) rises, which leads to an increase in Test Application Time (TAT) as well as the power requirement. This paper proposes a Fault-based Approximate Testing (FAT) technique that generates test patterns only for distinguished faults. The basic idea is identifying a set of faults (tolerable faults) that can be ignored and left untested. The remaining faults (intolerable faults) are tested by generating test patterns for those faults. We examine the impact of overlooking some of the faults by injecting stuck-at faults in the appropriate place of the netlist. Deviation in the output is analyzed in respect of an application-specific error threshold. The experimental result shows that irrespective of the erroneous result produced, these circuits can be used in all kinds of error-tolerant applications like image/video processing, speech processing, and digital communications. Along with the above benefit number of fault-sites is reduces by 40–50%, which leads to a reduction in the number of test patterns.
1 January 2022 in International Journal of Swarm Intelligence
Digital watermarking is nowadays an active research area for protecting multimedia content such as authentication and copyright safeguard. The major challenges faced in most watermarking schemes ar...