Resource Allocation in Cloud-Based Distributed Cameras

The data collected from millions of public cameras needs to be retrieved, stored, and analyzed. A system is required which needs to allocate significant amounts of resources to analyze large-scale visual data. Cloud computing provides shared storage, computation, and various services to handle such a tremendous amount of data collected from distributed cameras. In order to reduce the overall cost of analysis, this paper presents a resource allocation algorithm that provides cost-effective resources with the degree of demand; scaling automatically in proportion to demand fluctuation. In the cloud, the users prefer reliable resources at minimum cost whereas service providers prefer efficient resources utilization with maximum profit. Hence, it is necessary to have resource bargaining that assures and satisfies both cloud users and service providers. We propose a bargaining scheme using a game theoretic approach to managing cost and resource utilization in cloud-based distributed cameras. Our experiments show that our approach can lead to 10-15% reduction in cost by dynamically utilizing the resources and switching the service provider when it gets a better deal from other service providers.