Performance of data transmission over a Gaussian channel with dispersion

Data transmission over a channel with additive white Gaussian noise (AWGN) is perhaps the simplest yet most fundamental data transmission scenario in wireless networks. Surprisingly, this scenario in providing quality of service (QoS) guarantees to data traffic has not been well studied, particularly when the channel capacity is dispersed due to finite coding length. In this paper, we investigate the impact of physical layer channel coding on the data-link layer QoS with focus on how the length of physical layer codewords influences the upper layer QoS performance. To achieve this goal, we propose traffic and service models that take into account the finite coding length nature at the physical layer. Then, based on the stochastic network calculus theory, we derive probability delay and backlog bounds for the data-link layer. The obtained analytical results are validated by comparing with simulation results. These results also demonstrate that our cross-layer analysis can effectively characterize the impacts of channel coding at the physical layer on the QoS performance at the data-link layer. To the best of our knowledge, this is the first work taking coding length effect in cross-layer QoS analysis. Since in practice, no channel coding has infinite codeword length, which is often assumed in the related literature, we believe our work sheds new light on cross-layer QoS analysis.