A Fading-Insensitive Performance Metric for a Unified Link Quality Model

Link quality model is widely used in system evaluations to simplify the simulation complexity. It is also important in practical systems for improving the accuracy of the link adaptation and the efficiency of radio-resource-management. Conventional linear average SNR characterization of fading channel performance lacks generality, since the same linear SNR value may lead to drastic block error rate differences in various fading channels. A unified metric is proposed in this paper for link performance characterization and quality modeling. This paper proposes a mutual-information-based (MI-based) link quality model, which contains separate modulation and coding models. The modulation model maps the received SNR to the mutual information symbol by symbol. The coding model maps the sum or average of the mutual information to decoding performance for each coding block. The existing methods based on the effective signal-to-noise-ratio (SNR) of a multi-state channel have limited accuracy in the mixed modulation cases. Compared with the existing models, the MI-model is simpler and easier to apply to mixed-modulation cases and different H-ARQ schemes. The simulation results verify the accuracy in different multi-state channels and give the comparison with the exponential effective-SNR-mapping (EESM) model.