On efficient error diagnosis of digital circuits

Abstract

The rising trend in large scale integration and design complexity has greatly increased the need f o r eficient design error diagnosis. W e present techniques for fast and efficient error diagnosis of digital circuits by eliminating to a large extent the set of false candidates identified by the diagnosis. The elimination of false candidate regions is conducted via distinguishing x's, flipping of values at the output of candidate regions, and combination of these techniques. Our algorithms help to improve both the speed and resolution of error diagnosis. Experimental results on combinational benchmark circuits showed that up to 92% improvement in diagnostic resolution and 74 96 speedup over the original region-based diagnosis can be achieved with our approaches.

Multiple-Fault Diagnosis Using Faulty-Region Identification

“…Note that the average fault diameter is 4.35 in Table IV, which is larger than that of the modeled region used in any published result of the region-based diagnosis for ISCAS benchmark circuits, such as [9] [10] [11]. Unlike the region-based diagnosis whose runtime increases exponentially with the increase of the modeled region's size, the proposed diagnosis framework requires no assumption on the size of the suspect region and its runtime is indifferent to the fault diameter.…”

“…One type of multiple-fault-diagnosis methods are region-based methods [9] [10] [11], in which all real defects are assumed to locate within a neighborhood. The region-based methods utilize the X (unknown) model to approximately represent the possible faulty behaviors of a defective region in the simulation and then rank the defective regions based on how their X-simulation results can match the failing responses.…”

“…In addition, their computation complexity increases exponentially with the increase of region's radius. The experimental results reported in [9] [10] [11] are limited to a small radius (such as 1). [12] proposes the concept of SLAT (single-location-at-a-time) patterns, which can generate a faulty response fully explained by a single stuck-at fault.…”

Multiple defect diagnosis using no assumptions on failing pattern characteristics

“…The X values are allowed to propagate to observable points. The use of X values in this proposed approach is different from previous diagnosis methods [3, 11,12,13], where X values are used to model the unknown behavior of a logically-bounded defect. However, in this proposed approach, X values are used to represent the simultaneous interactions among all potential sites.…”

Incremental diagnosis of multiple open-interconnects

“…Logic unknown has been used in region-based diagnosis as an alternative to fault models [2,14]. Unlike [2,14], this work tailors the notion of logic unknown to an incremental framework in which faults in random error locations, not in regions, are identified. The second phase reduces the number of candidate fault tuples through a generalized fault-simulation scheme that enumerates the effects of an arbitrarily complex fault.…”