Stochastic computer network under accuracy rate constraint from QoS viewpoint

In the past, the main concerns for competitive enterprise environment of computer networks were data transmission delays and costs. However, from the viewpoint of quality of service (QoS), internet service providers and enterprise customers are increasingly focusing on data accuracy during transmission. This study extends a stochastic computer network (SCN) to evaluate the system reliability, where the system reliability is the probability that demand can be transmitted through the SCN under the total accuracy rate. In such an SCN, each arc has several capacities, and an accuracy rate. We can regard the system reliability as a measurable index for assessing the SCN performance. A lower boundary point for (d, K) is a minimal capacity vector, which allows the system to transmit demand d under the total accuracy rate K. Using the minimal paths, we propose an efficient algorithm to find all lower boundary points for (d, K). The system reliability can then be computed in terms of all lower boundary points for (d, K) by the recursive sum of disjoint products (RSDP) algorithm. Several numerical examples are demonstrated to illustrate the utility, efficiency, and effectiveness of the proposed algorithm.