Percolation and blind spots in complex networks

Recent works on network security have focused on whether a complex network can maintain its integrability under attack or random node failures. In applications of increasing importance such as sensor networks, a somewhat different problem, namely, the occurrence of isolated nodes (or blind spots), is of great interest. We show that, for networks with a stronger ability to form global spanning clusters, it is relatively more difficult to eliminate blind spots, and vice versa. We use the framework of percolation to investigate this phenomenon. Our analysis also yields a formula for the average number of blind spots, which provide an explanation for several numerical findings.