VECOS: A Vehicular Connection Steering Protocol

Abstract

Due to their worldwide deployment, 3GPP mobile networks, particularly Long-Term Evolution (LTE), are gaining a lot of momentum, as are LTE-connected vehicles. While one may envision an LTE-connected vehicle as a nicely designed vehicle with sophisticated equipment, a conventional vehicle with a person using an LTE-enabled smartphone or tablet on board can be logically qualified for an LTE-connected vehicle. Maintaining an acceptable quality of service (QoS)/quality of experience (QoE) of LTE services for a user on board a moving vehicle is a challenging problem. One approach for that is to anticipate QoS/QoE degradation and to exploit the different radio access technologies, such as WiFi, that may be available at an LTE-connected vehicle or, in general, at an LTE-enabled user equipment (UE) on board the vehicle. For this purpose, this paper introduces a complete framework that proactively defines QoS/QoE-aware policies for LTE-connected vehicles (UE devices) to select the most adequate radio access out of the available access technologies (e.g., WiFi and LTE) that maximizes QoE throughout the mobility path. The policies are communicated to the users following 3GPP standards and are enforced by the UE devices. They take into account the service type, the mobility feature, and the traffic dynamics over the backhauls of the different available accesses. Two different models were proposed to model the network selection process. The first model is based on multiple-attribute decision making (MADM) techniques, whereas the second model is based on the Markov decision process (MDP). Moreover, the network selection process is modeled using a time-continuous Markov chain, and the performance of the proposed framework (VECOS) is extensively evaluated through NS2-based simulations considering the case of two wireless access technologies, namely, WiFi and cellular networks. The obtained results illustrate that in comparison with conventional vertical handoff mechanisms whereby WiFi is always selected whenever it becomes available, the proposed framework ensures better QoS and achieves better QoE throughout the time of the received service and the mobility path of the user, even in the case of errors in the prediction of the user's mobility.