Reliability measure based on average loss of capacity

We present an effective measure of system reliability based on average loss of call capacity that is appropriate for telecommunication hardware. We are led to consider a system consisting of units with many states with appropriately chosen performance effectiveness indices, while all atomic units are binary. In this respect, this effective measure of partial system availability is a variant of customer-centered reliability methodology. We also identify the importance of a new structure called a ‘pool’ that behaves as a hybrid of series and parallel configurations. We present an algorithm for approximately calculating partial system unavailability, which solves the system decomposition problem for arbitrarily complex highly reliable systems made of hierarchically arranged k-out-of-n sub-structures. The algorithm agrees with our intuition for important special cases and has been programmed into an object oriented software tool. This method has been successfully used for the reliability analysis of Globalstar™ gateways.