A centralized trust-based collusion attack strategy in conjunction with an integer linear programming is proposed to compromise the decision of the fusion center in a two-phase energy efficient collaborative spectrum sensing (CSS) strategy, namely EE-CSS, which was discussed in [1]. The proposed strategy aims to attack only when it is likely to compromise the decision of the fusion center (FC). The results show that the proposed collusion attack can impact the global false alarm and miss detection probabilities drastically. A mitigating strategy based on cross-correlation of sensing reports is proposed to identify the secondary users with unusual behavior and to eliminate them from the decision making process at the FC.