Traditional multimedia scheduling approaches assumed perfect control channels where each node has access to the knowledge of its neighbors. However, in practice the control channels are always constrained and nodes can only exchange limited information with their neighbors. In this paper, we investigate how imperfect neighbor information affects the multimedia scheduling. First, we formulate the optimal multimedia scheduling problem with the constraints on network information. Specifically, a constrained factor is introduced to capture the profile of control channels. Then, we consider two cases of the constrained factor distribution: 1) the class with finite mean and variance, and 2) a general class that does not employ any parametric representation. In each case, we investigate the relationship between the control gain and scheduling performance based on available network and multimedia information. We show that the control gain can be chosen properly such that the optimal distributed multimedia scheduling can be achieved with an exponential convergence rate. In addition, an explicit equation for asymptotic convergence rate is derived for each case. Finally, we use computer simulations to verify the analytical results.