This paper proposes an object-level rate control algorithm to jointly controlling the bit rates of multiple video objects. Utilizing non cooperative game theory, the proposed rate control algorithm mimics the behaviours of players representing video objects. Each player competes for available bits to optimize its visual quality. The algorithm finds an "optimal solution" in that it conforms to the mixed strategy Nash equilibrium, which is the probability distribution of the actions carried by the players that maximizes their expected payoffs (the number of bits).
The game is played iteratively, and the expected payoff of each play is accumulated. The game terminates when all of the available bits for the specific time instant have been distributed to video object planes (VOPs). The advantage of the proposed scheme is that the bidding objects divide the bits among themselves automatically and fairly, according to their encoding complexity, and with an overall solution that is strategically optimal under the given circumstances. To minimize buffer fluctuation and avoid buffer overflow and underflow, a proportional-integral-derivative (PID) control based buffer policy is utilized.
Digital video compression is a multifaceted process that continues to spur fresh research challenges. A video encoder encompasses myriad techniques, taking advantage of temporal, spatial and statistical redundancies present in video data. The main theoretical basis of most of the existing rate control algorithms is the classical rate distortion, which quintessentially defines a non linear relationship between the distortion and bit rate. The model allows a video encoder to allocate bits to the compressed video so as to minimize the distortion function given a bit rate constraint; normally, the higher the bit rate, the lower the distortion, and vice versa.
Effective object-based rate control can be designed for bit rate allocation among video objects subject to bit rate constraints, along with allocation schemes that can maximize the coding efficiency, the quality of individual audio visual objects and the overall bit rate accuracy. In this paper, we propose a joint multiple video object rate control algorithm that distributes bits to video objects using non cooperative game theory. The algorithm is based on the mixed strategy Nash equilibrium of a non cooperative bi-matrix game.
In this paper, we propose a non cooperative game theory based algorithm for VOP-level rate control. In the algorithm, each VOP is a rational player. All players compete with each other in a non cooperative game, to secure resources (bits) in order to maximize their profit (visual quality). The game is played iteratively, and the expected payoff of each play is accumulated. The game terminates when all of the available bits for the specific time instant have been distributed to VOPs. The advantage of the proposed scheme is that the bidding objects divide the bits among themselves automatically and fairly, according to their encoding complexity, and with an overall solution that is strategically optimal under the given circumstances.
Network and Bottleneck Router
Consider N video sessions sharing a bottleneck router. The video flows are transmitted over a wired network, possibly with a wireless last hop, where the video is transmitted to client from a wireless access point. We assume the UDP transport protocol for video data. For streaming videos, it is highly important to be adaptive to network status. Otherwise, either a large portion of video data will be dropped under congestion, or the available network bandwidth will be wasted. We assume the video encoder on the server side can adjust its bit rate R according to feedback from the receiver to react to congestion in the network.
We assume a bottleneck link where N video sessions share the link bandwidth. The bottleneck could be a router link along the path, or the last-hop wireless link, as shown in Figure. Loss of video packets occurs at the bottleneck link due to congestion and at the wireless link due to interference or collisions. Bottleneck router is assumed to enforce a drop probability p(t) when congestion occurs, and is p(t) determined according to an AQM algorithm, i.e., random early detection (RED).