I. Introduction
Vehicular ad hoc network (VANET) is a promising technology for data exchange on the road, where traffic related information can be exchanged for traffic management, and entertainment-related data can be transmitted to improve user experiences [1]. In VANETs, vehicles are allowed to transmit signals autonomously to other vehicles in broadcasting, multicasting or unicasting manner, referred to as vehicle-to-vehicle (V2V) communications [2]. Moreover, the basic awareness messages should be periodically broadcast by vehicles including position, speed, etc. [3]. Currently, multiple modes and heterogeneous vehicular network have been investigated for V2V communications, including dedicated short-range communications (DSRC) based on IEEE 802.11p and LTE-V2X [4]–[8]. In DSRC, a vehicle could transmit packets in a dedicated frequency band if the channel is sensed idle for a specific period and when the packet transmission collides with other vehicles’ transmissions, the vehicle should wait for a random backoff interval before re-transmission, as specified by the carrier sense multiple access with collision avoidance (CSMA/CA) protocol [5]. In LTE-V2X, time-frequency resources are structured as that in LTE and can be allocated to different vehicles for V2V communications by a base station globally or by vehicles locally. Particularly, in LTE-V2X mode 4, vehicles use the sensing-based semi-persistent scheduling (SPS) scheme to select and reserve resource blocks (RBs) for packet transmissions locally, which facilitates reducing vehicles’ reliance on base station (BS) [3]. Hence, LTE-V2X mode 4 is a promising mode in LTE-V2X to support V2V safety applications which cannot completely depend on the coverage availability and the scheduling of BS [6]. The performance of DSRC and LTE-V2X mode 4 has been analysed in [3], which is highly related to the vehicle density, transceiver-receiver distance and packet generation patterns. It was illustrated in [3] that there is no absolute superiority between those two modes.