Non-orthogonal V2X communications

Abstract

In a cooperative intelligent transportation system (C-ITS), sensor-assisted vehicles share the collected data with each other through vehicle-to-everything (V2X) communications. This enables high-level applications such as safety-related service and provides a high-quality driving experience. However, the increasing number of intelligent vehicles leads to serious aggregated interference and congested communications. This increases the transmission latency and degrades the quality of service (QoS). To provide low-latency and reliable communications, non-orthogonal multiple access (NOMA) is recently advocated for V2X communications. By allowing multiple terminals to share the same time and frequency channel resources, NOMA improves the spectral efficiency and reduce the transmission latency, especially for a high-dense network. However, the overlapping of signals from multiple users introduces new challenges to the receiver design. This thesis exploits NOMA operating in two basic V2X communication scenarios: information exchange in a two-way relay channel (TWRC) and message broadcasting. Two non-orthogonal technologies are investigated: physical-layer network coding (PNC) and multiuser detection (MUD). The major difference between PNC and MUD is the decoding objectives. PNC aims to decode the network-coded messages while MUD is to recover individual messages. First, we study a TWRC operating with orthogonal frequency-division multiplexing (OFDM) modulated PNC and propose an inter-carrier interference (ICI) aware approach that jointly achieves accurate channel estimation, signal detection, and channel decoding. In highly-mobile networks, the carrier frequency offsets (CFOs) due to high-speed motion will lead to ICI in OFDM systems. Moreover, the vehicular environment with time-frequency-selective channels further undermines accurate channel estimation for multiple users. It is also worth noting that the CFO that exists in OFDM-modulated PNC cannot be completely eliminated through CFO tracking and equalization as in conventional point-to-point transmissions. These critical issues can significantly increase the bit error rate (BER) at the receiver. To address these challenges, we first express the channel estimation, and detection and decoding as two optimization problems. Then the expectation maximization (EM) and the belief propagation (BP) algorithms are employed to resolve the two optimization problems, respectively. Simulation results verify that the proposed approach can efficiently mitigate the negative effect of ICI by exploiting both pilot and data tones in channel estimation, detection, and decoding. The proposed ICI-aware approach is further implemented with software-defined radio (SDR). Experiment along a road-side environment was conducted on the campus. The V2X trial involved two types of nodes: terminals, which can be either vehicle or pedestrian, that aim to exchange self-information with each other, and a relay that is played by a road side unit (RSU). Each node was equipped with a universal software radio peripheral (USRP) platform for data transmission. The empirical results further verified that the proposed ICI-aware approach provides low BER and low latency compared with the conventional algorithm. Besides, the proposed scheme can decode both network-coded and individual messages. An interesting finding is that when the receiver fails to decode the network-coded messages, it is of strong possibility to recover the individual messages. This provides insights into our work on the MUD. Broadcasting is very common in V2X communications. For instance, it involves basic safety messages (BSMs) for safety applications and multimedia content for entertainment services. In urban areas, dense buildings obstruct the radio channel between vehicles in different road segments and non-line-of­sight (NLOS) propagation lowers the received power level. Therefore, messages that originate from adjacent roads (traveling in non-parallel directions) usually have weaker signals as compared with messages generated from vehicles on the opposite road segment (traveling in parallel directions). Besides, the high vehicular density leads to insufficient orthogonal channel resource and high interference. In this thesis, we propose to apply NOMA for V2X communications at the road intersection to enhance the spectrum efficiency and improve the package delivery ratio (PDR) performance. With NOMA, signals from more than one transmitters are decoded together and the interference from undesired users can be canceled once its message is obtained. This thesis studies two NOMA-based V2X communication schemes, namely, NOMA-V2X decoded by successive interference cancellation (SIC-V2X) and NOMA-V2X decoded by joint decoding (JD-V2X). Based on the tools developed in stochastic geometry, we derive and compare the PDR expressions for both NOMA schemes and the orthogonal multiple access (OMA) scheme. The results indicate that 1) both NOMA schemes outperform the conventional OMA scheme and the PDR of LOS/NLOS communications with two-user access increases by 51%/369%; 2) for four-user access, the proposed NOMA scheme shows 375% goodput enhancement as compared with the OMA scheme; and 3) the JD-V2X provides significant PDR enhancement compared with SIC-V2X in the high data rate regime. Besides the tractable network model to analyze C-V2X broadcast performance operated with NOMA, the feasibility of applying NOMA to C-V2X communications is studied. Focusing on the PC5 sidelink interface that is enabled in the 3GPP Release 14 for direct communications, we discuss the implementation of NOMA in the radio interface and propose two NOMA receivers based on SIC and JD techniques. Compared with the conventional C-V2X receiver operated with OMA, modifications are needed at the channel estimator, equalizer, and demodulator, as well as the inclusion of an interference canceler so that multiple users can be decoded from the overlapping signals. Besides, the proposed NOMA receivers support both single-antenna and multi-antenna systems, and the impact of the number of receiving antennas on the block error rate (BLER) is studied. Overall, this thesis investigates non-orthogonal V2X communications. The research combines the concepts of NOMA and network coding to address the basic issues and new challenges from the increasingly high-dense vehicular networks. The algorithms proposed in this thesis are expected to provide insights into the development of ultra-reliable and low-latency V2X communications in future smart cities.