Adaptive Generative Models for Digital Wireless Channels

Generative models, which can generate bursty error sequences with similar burst error statistics to those of descriptive models, have an immense impact on the wireless communications industry as they can significantly reduce the computational time of simulating wireless communication links. Adaptive generative models aim to produce any error sequences with any given signal- to-noise ratios (SNRs) by using only two reference error sequences obtained from a reference transmission system with two different SNRs. Compared with traditional generative models, this adaptive technique can further considerably reduce the computational load of generating new error sequences as there is no need to simulate the whole reference transmission system again. In this paper, reference error sequences are provided by computer simulations of a long term evolution (LTE) system. Adaptive generative models are developed from three widely used generative models, namely, the simplified Fritchman model (SFM), the Baum-Welch based hidden Markov model (BWHMM), and the deterministic process based generative model (DPBGM). It is demonstrated that the adaptive DPBGM can provide accurate burst error statistics and bit error rate (BER) performance of the LTE system, while the adaptive SFM and adaptive BWHMM fail to do so.