Collecting statistics from client requests, the broadcast server partitions data items into hot and cold-item sets with the optimal cut-off point. Hot items are broadcast periodically with periods based on their access probabilities. In a time slot with no hot items scheduled, the server broadcasts a proper cold item considering the waiting time and the number of outstanding requests. We analyze the optimal the cut-off point by calculating average response time as a function of the cut-off point. Simulation results show that our proposed algorithm outperforms existing methods in various circumstances.