A Software Defined Multi-Path Traffic Offloading System for Heterogeneous LTE-WiFi Networks

Ever growing mobile users, devices and applications lead to tremendous volume of mobile network traffic nowadays. Current cellular network infrastructure is facing the problem of lacking sufficient bandwidth. Many existing works propose to solve this issue by offloading mobile traffic to wireless local area network. However, traditional single path communications limit the traffic offloading performance due to the network handoff caused by user mobility. To further explore the benefits and potential of traffic offloading in heterogeneous mobile networks, we develop a software-defined and multipath-based network aggregation system specifically for LTE-WiFi traffic offloading in this paper. Our system relies on Multipath TCP and Software-Defined Networking technologies: the former is run on mobile devices to create TCP subflows over multiple network connections, while the latter is leveraged to direct those subflows for proper usage of network bandwidth via a centralized controller. To obtain the appropriate bandwidth allocation, we formulate traffic offloading as a network optimization problem. Furthermore, the controller configures the rate of subflows in each radio access network to optimize the objective of load balancing and fairness. The devised controller is also responsive to network status change, and the centralized design facilitates network resource management and policy making. A prototype of our traffic offloading system is implemented using network emulator Mininet-WiFi. Emulation results prove its efficiency in throughput maximization, fairness, and service continuity in heterogeneous LTE-WiFi networks.