Constructing a Green MPTCP Framework for Industrial Internet of Things Applications

Abstract

In the typical distributed applications, the data exchange between the communicating peers proceeds along the transport path established in the connection initialization phase, even if a better one is discovered during the active session. With the recent advancement in the multipath protocol development, e.g., MPTCP, the peers can benefit from a concurrent use of a few channels, thus improving the transmission quality. However, the present approaches to the multipath transfer organization tend to neglect the energy aspects, crucial for resource-constrained Internet of Things (IoT) devices. In this paper, a framework for MPTCP module tuning, targeting the power expenditure, is developed. A new Scheduler and a new Path Manager promoting a conservative energy economy are designed by adopting a formal optimization approach. Moreover, explicit guidelines regarding the TCP variant selection are provided. As confirmed by numerous experiments involving physical devices and real networks, the proposed configuration scheme allows for several percent energy gain with respect to the default one, thus setting a solid framework for green MPTCP-based Industrial IoT communication.