Ultra-Reliable Multi-Connectivity With Negatively Dependent Fading Channels

Ultra-reliable wireless communications for critical industrial infrastructure communications is challenging because of the random fading channel. In order to improve the outage performance, diversity techniques are studied extensively and are currently standardized in 5G and beyond. The dependencies of fading channels originate from the random propagation environments, local scatterers, multipath conditions, and shadowing effects. It is well known that spatial and spectral correlation deteriorates the outage performance and well accepted that independent fading channels provide the highest reliability gains. In this work, we consider a negative dependence between the fading channels and show that the outage performance can be significantly improved by negative dependencies. To this end, we construct an arbitrarily dependent bivariate exponential model and show that it supports the whole range of dependency. Based on this construction, we show that the zero-outage capacity is positive for the completely negatively dependent case even under short-term power constraints, i.e., both link selection and link combining achieve zero-outage performance. Furthermore, we show the monotonicity of the outage probability for the relevant operating regime. Finally, we completely characterize the impact of correlation on the average performance in a bivariate binary channel model. Numerical assessments confirm the closed-form expressions of the outage probability and motivates to deploy network nodes in order to create negatively dependent fading channel gains.