IEEE Asilomar Conference on Signals, Systems & Computers, pp. 131-135, October 26th - 29th, 2008. (invited paper)
Spatial interference cancelation based on zero-forcing beamforming significantly improves the throughput of a mobile ad hoc network. Nevertheless, the performance of spatial interference cancelation can be degraded by imperfect channel estimation at receivers. This paper analyzes the impact of inaccurate channel state information (CSI) on network throughput by modeling transmitters as a Poisson point process and applying mathematical tools from stochastic geometry. An algorithm is proposed for opportunistically estimating CSI required for interference cancelation. Based on this algorithm, the distribution of residual interference is derived, which exists at each receiver due to imperfect CSI. In particular, interference power is shown to be linearly proportional to the transmitter density. These results facilitate the analysis of network transmission capacity, defined as the maximum transmitting node density under constraints on outage and the signal-to-interference-noise ratio. Simulation results show that spatial interference cancelation increases transmission capacity by an order of magnitude or more despite CSI inaccuracy. Furthermore, capacity loss due imperfect CSI diminishes as the transmitter density decreases.