This paper describes a novel scheme which enables scaling of cellular node applications without disrupting session continuity. With this scheme, states of node applications, including User Equipment (UE) states and transport protocol states, are maintained during scaling operations that are possible in a virtualized environment. A prototype of Mobility Management Entity (MME) of Evolved Packet Core (EPC) was designed to show how the proposed scheme together with the Elastic Core Architecture (ECA) can achieve the session continuity by means of empirical evaluations performed in a laboratory test bed. The results show that our proposed scheme can realize virtualization-based scaling of stateful cellular network node applications with negligible session interruption characteristics.

To implement cutting-edge services such as highresolution movie streaming and smart metering on mobile networks cost-effectively, mobile operators must meet various conflicting requirements. For example, they must manage a massive number of devices in smart-meter-type machine-to-machine (M2M) applications for which quality-of-service (QoS) requirements are quite lenient. At the same time, they need to support stringent QoS requirements in the form of a few milliseconds of delay and a high bandwidth guarantee for a smaller number of devices in video and remote surgery applications. As building a dedicated per-service physical network is very costly, network slicing by means of resource virtualization was developed to accommodate such heterogeneous services in a single physical network infrastructure. In network slicing, network resources are isolated to form slices, which then can be used to provide different services. This slicing is helpful in accommodating conflicting, often incompatible services at the cost of losing the multiplexing gain achieved in monolithic composite service networks. The loss in multiplexing gain is not a big problem when the number of slices is relatively low. However, numerous services are provided in cellular networks, and the number is expected to be drastically higher in the 5G era. Creating perservice slices would lead to wastage of resources. In this regard, we aim at balancing the need for isolation to meet conflicting service requirements against resource usage efficiency. We investigated possible means of simultaneously achieving isolation and minimizing the loss of multiplexing gain. Our findings will aid in the development of network management architecture suitable for the 5G era and beyond.

No abstract

No abstract

No abstract