Applying Game Theory to Rate Control Optimization for Hierarchical B-Pictures

Abstract

The hierarchical B-picture structure has been widely applied for its scalability and coding efficiency. The bit allocation problem among frames in different temporal layers can be modeled as a multi-objective optimization among multiple players, which is similar to the bargaining problem. In this paper, we propose a bargaining game theory based frame-level bit allocation and rate control algorithm for hierarchical B-picture structure. First, the bit allocation problem among frames that belong to different temporal layers is modeled as a bargaining game, according to the efficiency and fairness requirement of sharing the available bandwidth resources. Second, the dependency relationship between different temporal layers and linear rate distortion model are employed to define the utility function in the bargaining game. Finally, the bandwidth is allocated for the frames based on the nash bargaining solution. Experimental results demonstrate that the proposed algorithm achieves better performance on bit rate accuracy, coding efficiency and fairness, as compared to other benchmark one-pass rate control algorithms.