An ICI-Aware Approach for Physical-Layer Network Coding in Time-Frequency-Selective Vehicular Channels

摘要

Applying physical-layer network coding (PNC) to vehicular ad-hoc networks (VANETs) can theoretically boost the network throughput by 100%, thus partially addressing the intermittent node connectivity and short contact time issues caused by high speed vehicle motions. However, the application of OFDM modulated PNC in VANETs faces detrimental effects caused by carrier frequency offsets (CFOs) and time-frequency-selective channels. CFOs may destroy the orthogonality of OFDM subcarriers, resulting in inter-carrier interference (ICI). The CFOs of two transmitters may also be different, and cannot be removed by CFO tracking and equalization at the receiver as in conventional single-user communication even if the CFOs are known. In addition, time-frequency- selective channels due to delay and Doppler spreads are difficult to estimate and non-accurate channel estimations will increase the detection bit error rate (BER). To address the two challenges, this paper proposes an ICI-aware approach that jointly exploits pilot and data for channel estimation and data detection. Specifically, our approach jointly uses the belief propagation (BP) algorithm to mitigate the CFO/ICI effect for data detection, and the expectation maximization (EM) algorithm to accurately estimate the channels. A linear interpolation method and an ICI compensation method are simulated as benchmarks. Simulation results indicate that our approach improves the BER performance compared to the two benchmarks (more than 2 dB SNR gain in most cases), especially in the high SNR regime.