EVA: Enhancing VoIP applications

This paper introduces EVA, a mechanism that enhances VoIP applications by optimizing path selection, taking into consideration multihoming, traffic performance and application performance criteria. These criteria are hierarchically organized according to their type, which can be classified as benefits or costs. EVA solves the NP-hard path selection problem through a MADM technique that employs a new distance function and a stability factor to select paths with optimal quality, while assuring the highest quality of experience during the VoIP call. The EVA solution mitigates unstable scoring, commonly found in MADM techniques, while keeping the same complexity O(m·n). EVA is evaluated analytically using data from a wireless testbed with multihomed hosts. Results demonstrate that EVA selects paths with high quality of experience levels and it is able to maintain the quality levels throughout the sessions. Moreover, when compared to similar techniques, the proposed approach demonstrates higher flexibility and adaptability to heterogeneous scenarios with different characteristics.