Abstract
In this paper, we derive tight bounds for the eavesdropping attacks on continuous variable quantum key distribution (CV-QKD) protocol that involves nonmaximally entangled states. We show that deriving bounds on the eavesdropper's accessible information based on the Heisenberg uncertainty yields upper bounds, but those bounds are not tight. For this reason, we follow different techniques to derive the desired tight bounds. The new bounds are tight for all CV-QKD protocols that involve two-mode entangled state. Our derivations are applied to direct and reverse reconciliation schemes of protocol implementation, respectively.