Sebastian Schaal

Constrained-random generation is the main method used for test-case automation for pre-silicon digital
block verification. Constrained-random generator and simulator performance means that it is usually possible to close coverage using this method on a small or medium-sized block within a week or a month with little manual intervention. However, the redundancy inherent in constrained-random generation means that it can take many months and considerable manual effort to achieve 100%… Mehr anzeigen

Constrained-random generation is the main method used for test-case automation for pre-silicon digital
block verification. Constrained-random generator and simulator performance means that it is usually possible to close coverage using this method on a small or medium-sized block within a week or a month with little manual intervention. However, the redundancy inherent in constrained-random generation means that it can take many months and considerable manual effort to achieve 100% functional coverage for large and complex blocks.
Unsupervised machine learning techniques, such as anomaly detection, look for unusual or novel behaviour. This raises the possibility of using such techniques to reduce the redundancy in simulated tests. Here we report positive results in closing coverage on a large and complex block using these techniques. The results suggest that for this block the time for coverage closure can be reduced from six months to less than three months. This still allows for considerable soak testing and, moreover, the tests used for soak testing will be different from each other. As a result of this work the Infineon Automotive Microcontroller Division is using machine learning in the verification of complex designs. We believe the techniques described here have real potential for wider practical application. Weniger anzeigen

Numerous authentications take place in our everyday lives: From the use of passwords to access an email account, fingerprint scans to grant access to buildings to the confirmation of a product's origin to prevent counterfeits. Sensor technology at the same time has advanced tremendously over the last decades. Not only are sensors making things increasingly smart (e.g. smart homes, smart cars), sensors are also increasingly included in consumer products, such as cell phones or digital… Mehr anzeigen

Numerous authentications take place in our everyday lives: From the use of passwords to access an email account, fingerprint scans to grant access to buildings to the confirmation of a product's origin to prevent counterfeits. Sensor technology at the same time has advanced tremendously over the last decades. Not only are sensors making things increasingly smart (e.g. smart homes, smart cars), sensors are also increasingly included in consumer products, such as cell phones or digital cameras.
The balance between security and convenience, however, is hard to find. This represents huge business opportunities for new business models. This report analyzes how sensor-based authentication may look like in the future, describes scenarios and presents potential business ideas.

ISBN: 978-3-9815538-5-7 Weniger anzeigen