Abstract
In this letter, we investigate the best antenna placement for eavesdroppers (Eves), who intend to wiretap information from the legitimate transmitter (Alice). Two configurations of co-located or distributed antennas (CA or DA) are compared to find the most harmful eavesdropper setting, which achieves higher average eavesdropping rate, i.e., the long-term leakage rate to Eves averaged over channel fading and random locations. For K → ∞, where K is the number of available antennas for Eves, we prove that without a guard zone (i.e., a geometric zone around Alice where Eves cannot be located in), DA Eves are more harmful, while with a nonzero guard zone, CA Eve is superior. For the finite-K case, we derive upper/lower bounds for the average eavesdropping rate. The bounds are then used to find a DA region, in which the DA Eves definitely outperform CA Eve. It is shown that this region depends on the value of K, the radius of the guard zone, and the jamming at Alice.