Application-specific slicing for MVNO and traffic characterization [Invited]

In this paper, we apply the concept of a software-defined data plane to defining new services for mobile virtual network operators (MVNOs) to empower them with the capability of tailoring fine-grained subscription plans that can meet users' demands. For this purpose, we have recently proposed the concept of application-and/ or device-specific slicing that classifies application-and/or device-specific traffic into slices and introduced various applications of our proposed system [Optical Fiber Communication Conf. (OFC), 2016, paper Th3I.4]. This paper reports the prototype implementation of the proposal into a real MVNO connecting customized smartphones so we can identify applications/devices from the given traffic with 100% accuracy. We first characterize the traffic patterns of popular applications to build a detailed understanding of how network resources are utilized for real users and applications and what their demands on different network resources are. Then, we classify traffic according to the devices and find that the flow characteristics of LTE RAN slices are very similar to those of 3G. We discover that most of the bandwidth is consumed by those flows with the duration of several hundreds of seconds, and the flow size is larger than 5 Mbytes in both LTE and 3G MVNO networks.