Engineering Machine-to-Machine Traffic in 5G

Abstract

Machine to machine (M2M) traffic is characterized as low-rate, small-packet traffic with correlated transmissions. In this paper, we propose a two-phase traffic control mechanism (2PTC) for carrying M2M traffic in the future fifth generation (5G) networks. In this mechanism, the packets from each machine are directed to a virtual serving gateway associated with the machine, which receives and aggregates traffic from multiple machines and forwards the aggregate traffic to the sink. The first communication phase takes place through a simple single-path routing technique, while the second phase is empowered by multipath traffic engineering (TE) optimization. At the virtual serving gateways, traffic aggregation (TA) may include network-layer flow trunking and application-layer content compression. At the core of 2PTC is joint gateway selection and machine-to-gateway association that favors TA potentials while minimizing association cost and virtual serving gateway count. The described problem is formulated as a mixed integer programming optimization problem. As the structure of the formulation is inherently NP hard, the problem is solved using relaxation and rounding techniques, whose solution quality is evaluated through numerical analysis. We also implement the solution in a network simulator and evaluate the performance of 2PTC, through extensive simulations. Simulation results indicate that routing M2M traffic to properly selected virtual serving gateways for TA can alleviate the large-quantity small-packet problem, while enhancing the performance of the background traffic. This paves the way for further performance enhancement or enabling new features when deploying virtual network functions at virtual serving gateways.