In order to enhance transaction security in wireless sensor networks, it is important to evaluate nodes? trustworthiness. Malicious nodes may strategically alter their behavior for concealing malicious behavior and prompting their reputation. Collusive or malicious behaviors might change the system trust entropy due to their voting ratings are biased. In this paper, according to the basic theory of Shanon informational entropy, the trust entropy and standard structure entropy of distributed entities trustworthiness evaluation are proposed, which be used to detect whether the malicious behavior happened in sensor systems. By simulating experiment, it can detect whether malicious behavior happened where the system is attacked either maliciously, randomly or collusively. Especially in collusive attack, the malicious nodes will boost their clique reputation and drop honest nodes by bad mouthing attack.