Packet-level error models for digital wireless channels

Packet-level error models have great significance for the design and performance evaluation of higher layer wireless communication protocols. These models aim to characterize the statistical behavior of bursty packet error sequences encountered in digital wireless channels. In this paper, target packet error sequences are generated by computer simulations of a coded General Packet Radio Service (GPRS) system with a typical urban (TU) channel and co-channel interference. A general design procedure of a generative model is then proposed by using a properly parameterized and sampled deterministic process with a threshold detector and two parallel mappers. Simulation results indicate that the proposed deterministic process based generative model (DPBGM) allows us to approximate very well the characteristics of the target packet error sequences with respect to the gap distribution (GD), error-free run distribution (EFRD), error cluster distribution (ECD), error burst distribution (EBD), error-free burst distribution (EFBD), block error probability distribution (BEPD), and packet error correlation function (PECF). Importantly, it is shown that both the accuracy and the efficiency of the suggested DPBGM outperform those of a widely used Markov model.