HARQ过程条件错误概率分析

Conditional Error Probability Analysis in HARQ Process

在混合自动请求重传(HARQ)技术应用中,之前的传输失败是重传的前提,合并后数据包中的噪声已不再是零均值高斯分布,且与发射信号相关。然而,在目前HARQ及其他类似技术的分析中,普遍假设合并后的噪声仍服从零均值高斯分布,导致对错误概率估计偏小。该文对重传合并后的条件错误概率进行了精确分析,提出了条件错误概率的近似表达式,为进一步分析提供了方法,并比较了传统假设条件下的分析结果。在该假设条件下的错误概率估计与真实值的误差随数据包符号数的增加而减小。重传时信道增益越小,误差越大。

In hybrid automatic repeat request （HARQ） process, the failure of the first transmission is the condition of the retransmission. The distribution of the noise after combining is not zero-mean-Gaussian. Unfortunately, the assumption of zero-mean-Gaussian noise is widely used in the analysis of HARQ, as well as some similar technologies. Therefore, the estimation of the conditional error probability is smaller than the exact one. In this paper, the conditional error probability under non-zero-mean-Gaussian noise is exactly analyzed. An approximate expression of the exact conditional error probability is also presented for simplicity of application. The difference between the real error probability and estimation based on zero-mean-Gaussian noise assumption is more obvious if the number of symbols in a packet is small and the channel of retransmission is worse than that of the first transmission.