Decoupling network services from underlying hardware, network function virtualization (NFV) is expected to significantly improve agility and reduce network cost. However, network services, sequences of network functions, need to be processed in specific orders at specific types of virtual machines (VMs), which couples decisions of VMs on processing or routing network services. Built on a new stochastic dual gradient method, our approach suppresses the couplings, minimizes the time-average cost of NFV, stabilizes queues at VMs, and reduces the backlogs of unprocessed services through online learning and adaptation. Asymptotically optimal decisions are instantly generated at individual VMs, with a cost-delay tradeoff [(ϵ)/√ϵ]. Numerical results show that the proposed method is able to reduce the time-average cost of NFV by 30% and reduce the queue length (or delay) by 83%, as compared to existing non-stochastic approaches.
|Original language||English (US)|
|Title of host publication||2017 IEEE Global Communications Conference, GLOBECOM 2017 - Proceedings|
|Publisher||Institute of Electrical and Electronics Engineers Inc.|
|Number of pages||6|
|State||Published - Jul 1 2017|
|Event||2017 IEEE Global Communications Conference, GLOBECOM 2017 - Singapore, Singapore|
Duration: Dec 4 2017 → Dec 8 2017
|Name||2017 IEEE Global Communications Conference, GLOBECOM 2017 - Proceedings|
|Other||2017 IEEE Global Communications Conference, GLOBECOM 2017|
|Period||12/4/17 → 12/8/17|
Bibliographical noteFunding Information:
Work in this paper was supported by the National Natural Science Foundation of China grant 61671154, the Innovation Program of Shanghai Municipal Education Commission; and US NSF 1509005, 1508993, 1423316, 1442686, 1202135.