Hybrid scale-based approximation algorithm for QoS routing

With the widespread use of network applications (VoIP, IPTV, etc.) which are sensitive to multiple QoS parameters (delay, jitter, error rate, financial cost, etc.), multi-constrained QoS routing becomes increasingly desired. Such routing is proved to be NP-hard when the number of additive metrics to be optimized is equal or greater than two. Thus, multiple fully polynomial approximation algorithms have emerged. These algorithms are essentially based on two scales: linear and logarithmic. In this paper, we prove that both the two scales could be inefficient for QoS path pre-computations and then propose a new algorithm that employs a hybrid scale to enhance path computations. As we proved it formally and by simulations, our algorithm guarantees reduced time responses for the pre-computed QoS paths without deteriorating their approximation quality.