The paper proposes a testing environment based on Colored Petri Nets (CPN) for analyzing the behavior of Context- Oriented Software (COS). The token of CPN and the token game are suitable for representing the layer of COS and its (de- )activation respectively. Especially, we expect CPN can detect mutual exclusion faults related to the layer interactions. Even though, if CPN finds a fault on COS, we cannot easily find where is the cause of the fault in the software since actual environments are compounded. Thus, in a behavior model of COS, we cannot find easily which event fails in changing layers. In this paper, we focus on two issues that are called compound environments, and context events. To overcome these issues, firstly, we clarify the advantages of CPN to state machines. Secondly, the paper introduces a simulation testing environment based on these advantages. In this testing environment, the state space graph of CPN contributes to finding a cause why fault occurred.