Energy-Efficient Multimedia Data Dissemination in Vehicular Clouds: Stochastic-Reward-Nets-Based Coalition Game Approach

摘要

In this paper, we investigate an energy efficiency issue for multimedia applications in a vehicular cloud environment. The problem of energy efficiency is formulated as a stochastic reward nets (SRNs)-based coalition game, in which vehicles are assumed as the players that formulate the coalition among themselves using a predefined criteria based on the demand generated and the available resources at the nearest cloudlet. A demand- and supply-based payoff value function is formulated for each player in the game. The processes and actions of the players are represented as SRNs in which each player should have a finite number of tokens to fire its actions. To reduce the delay, each vehicle accesses resources such as memory, processing power, and storage from the nearest cloudlet that may be either at the road side units deployed near the road or on the vehicles. The actions of the players are associated with a cardinality that represents the number of rewards and actions they have taken in a unit interval of time after interacting with an environment that is stochastic in this paper. For each action performed by the players, they may get feedback in the form of a reward or a penalty, according to which each player updates its action probability vector that helps them take the next actions in the game. An energy-efficient algorithm for frame scheduling using the nearest cloudlet is also proposed. The performance of the proposed algorithm is found to be satisfactory with respect to various evaluation metrics. In particular, there is an increment of 10%–15% in profit generation, with 20%–25% reduction in the delay, and an increment of 20% in the packet delivery ratio.