This paper proposes a game-theoretic framework for analyzing the decentralized and centralized control of smart grids based on the availability of information. For the demand response, demand-side actors in smart grids need to obtain various types of information via communication, e.g., a house with a photovoltaic (PV) power system acts efficiently based on the weather forecasts. In contrast, the information required for control is not always available because of communication failure. If information is unavailable, other control methods can cope with loss of the precise information. This paper introduces a comprehensive framework for a demand side management system for PV systems. According to the availability of information to predict the amount of PV power generation, we evaluate three control schemes, i.e., decentralized open-loop control, decentralized feedback control, and centralized control. Two types of decentralized control are formulated using a differential game, whereas centralized control is formulated as an optimal control problem. Considering the output of a PV system, each demand-side actor schedules their power consumption to minimize a cost function, including the disutility, electricity rates, and the supply-demand balance. Simulation results reveal that decentralized open-loop control is useful when information about the predicted data of power generation is available, whereas decentralized feedback control works efficiently when information is unavailable.