An approach based on matching theory to distributed deployment of NFV-based network services to geo-distributed edge computing systems

• Main Authors: Chen, Yi-Chia, 陳奕佳
• Other Authors: Yen, Li-Hsing
• Format: Others
• Language: en_US
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/h7kasp

碩士 === 國立交通大學 === 資訊科學與工程研究所 === 107 === With both Network Function Virtualization (NFV) and Multi-access Edge Computing (MEC) environment, network service providers can deploy their NFV-based network services on the edge servers maintained by edge service providers (ESPs) to serve end users. Nevertheless, determining which ESP to lease resource and how much to pay for the deployment will become an important issue for the network service providers. For the ESPs, determining which network serive to serve and how much to charge from its provider is also an important issue. We proposed a two-layer mechanism of matching game for the deployment of network services to the ESPs. Specifically, the upper layer is the bargaining process between network services and ESPs, we use matching auction; the lower layer is resource allocation for VNFs within an ESP, we use one-to-many matching model. The proposed mechanism can provide a weakly stable result, that is, there will be no network service that is more favorable than all ESPs' current matching results. However, there may exist a network service which can have more profit if it change its current partner. We simulated the proposed mechanisms and showed that we can have a higher average number of served network services, and both ESPs and network service providers can have higher profits.