On the performance of successive interference cancellation in 5G small cell networks

The ideal successive interference cancellation paradigm helps to achieve the capacity of some multiuser channels, such as the Gaussian multiple access and broadcast channels. However, its performance is much more modest under realistic constraint on the decodability of the interference signal. In this paper, we rely on basic stochastic geometry models to analytically evaluate the performance of interference cancellation receivers in a local area network scenario, under `realistic' rate-constraints on the decodability of the interference signal. Analytical findings are validated by extensive Monte Carlo experiments. Alongside, complementary system level simulations results are presented to demonstrate the performance in a `practical'-like system. Our findings explicitly quantify how the gains from interference cancellation techniques depend on the spatial density of the active interferers in the network, and their respective data rates. The findings further highlight the importance of properly dimensioning the system in order to fully benefit from such interference cancellation techniques.