The Design of Inherently Fault-Tolerant Systems

With very large scale integrated (VLSI) circuits having as many as 10 million circuit components on a silicon chip and the potential for an order of magnitude more when considering wafer scale integration (WSI), two questions arise. First, how can all these components be organized so that a useful function is performed? For a device such as a random access memory, the organization is relatively obvious. However, if processing elements are considered, the organization is significantly complicated and may change, depending upon the application. Second, how can all these circuit components work together reliably? One approach, known as fault avoidance, is to manufacture the parts using very rigorous and expensive procedures and to take into consideration all possible environmental influences when designing, fabricating, and packaging the device. A second approach is to employ fault-tolerant design techniques.