This work addresses the problem of efficient virtual sensor formation in sensor-cloud for provisioning green Internet of Things (IoT)-based services. The sustainability of sensor-cloud primarily depends on its quality of service (QoS), resource utilization efficiency, and the revenue earned by its stakeholders. As sensor nodes are resource-constrained, it is important to minimize the overall energy consumption at these devices and for the overall network, thereby reducing the cost of service provisioning. Additionally, it is important to ensure that there is no compromise in the QoS to establish a continuous inflow of revenue for the service-provider. Existing works on virtual sensor formation in sensor-cloud barely consider resource consumption due to multihop communication from source nodes to the base-station, which plays a significant role in determining the resource consumption of the network. Thus, in this work, we model this problem using a game-theoretic approach and propose a hybrid algorithm, named GS2, to solve it. We observe that, by modifying the price paid to the sensor-owners, the service-provider is able to control the resource utilization of the network. Simulations show that GS2 reduces the overall energy consumption by 15.92-24.18%, and increases the revenues of the service-providers and sensor-owners by 8.33-18.33% and 21.05-38.09%, respectively.