Error Tolerant Compression studies the effect of faults in the multimedia compression systems, i.e. video, image and audio system. Classical manufacturing test for digital chips classifies chips into two categories: perfect and imperfect. Our work is motivated by the notion that in some instances imperfect chips can be used, as long as they introduce â€œacceptable errorsâ€?. Error tolerance (ET) leads to a relaxation of the requirement of 100% correctness for devices and interconnects, which may dramatically reduce costs for manufacturing, verification, and testing. Categorizing chips into acceptable and unacceptable also leads to increases in yield rate. Determining what constitutes acceptable degradation in system performance is obviously an applicationspecific decision; both performance criteria and acceptability thresholds are highly dependent on the application.
In this project we consider multimedia compression systems as a promising application area for our proposed ET concepts. This is because i) many multimedia compression systems are deployed in consumer devices, for which maintaining low costs is important, and ii) compression itself leads to a lossy representation of signals, so that the effect of system faults can be viewed as an additional source of â€œnoiseâ€? or representation error. As an example, we studied a complete MPEG video encoder indicates that several of its building blocks, in particular its Motion Estimation (ME) component, and discrete cosine transform (DCT) are such that some hardware faults lead to acceptable degradation at the system outputs.
This project is supported in part by the National Science Foundation under Grant No. 0428940.
- I. Chong and A. Ortega, "Hardware Testing For Error Tolerant Multimedia Compression based on Linear Transforms". In Proc. of IEEE International Symposium on Defect and Fault Tolerance in VLSI Systems (DFT), Monterey, October 2005.PDF format
- H. Chung and A. Ortega, "Analysis and testing for error tolerant motion estimation". In Proc. of IEEE International Symposium on Defect and Fault Tolerance in VLSI Systems (DFT), Monterey, October 2005.PDF format
- I. Chong, H. Y. Cheong, and A. Ortega, "New Quality Metric for Multimedia Compression Using Faulty Hardware". In Proc. of International Workshop on Video Processing and Quality Metrics for Consumer Electronics, Arizona, January 2006.PDF format
- H. Y. Cheong, I. Chong and A. Ortega, "Computation Error Tolerance in Motion Estimation algorithms". In Proc. of International Conference on Image Processing (ICIP06), Atlanta, October 2006. PDF format
- I. Chong and A. Ortega, "Dynamic Voltage Scaling Algorithm For Power Constrained Motion Estimation". In Proc. of International Conference on Acoustics, Speech, and Signal Processing (ICASSP07), Hawaii, April 2007. PDF format
- I. Chong and A. Ortega, "Power Efficient Motion Estimation Using Multiple Imprecise Metric Computation". In Proc. of International Conference on Multimedia & EXPO (ICME07), Beijing, July 2007. PDF format
- H. Y. Cheong and A. Ortega, "Distance Quantization Method for Fast Nearest Neighbor Search Computations with applications to Motion Estimation", Asilomar Conference on Signals, Systems and Computers, Pacific Grove, CA, November, 2007. PDF format
- H. Y. Cheong and A. Ortega, "Motion Estimation Performance Models with application to Hardware Error Tolerance", In Proceedings of Visual Communications and Image Processing (VCIP), 2007. PDF format
- Sample Images with various Pth value
Here are some images produced by JPEG codec which has a fault inside. Po is error probability which means probability that we can see visible error in the image due the fault in the system. The error occurs in the basis of each 8X8 block of image. So "Po=0.01" means that we can see visible error every 100th block of total 8X8 blocks of image. For example, 256X256 lena image has 1024 8X8 blocks. If Po=0.01 there can be around 10 blocks with visible error in this image. Here left image is error-free one, and right one is with error.