Toward a programmable software-defined IoT architecture for sensor service provision on demand

In the age of Internet of Things (IoT), sensors form a foundational component of IoT services, yet they are rigid with little capability for programmable configuration or reusability as they are application-specific, manufacturer-specific. Emerging IoT applications often deploy a vast number of sensors which may serve multiple applications. Programmability is thus essential but not found in legacy or current generation sensors. It is challenging to effectively utilize heterogeneity of resources to handle a large number of application demands. Software Defined Networking (SDN) and Network Function Virtualization (NFV) have proved effective paradigms for provisioning services on-demand and managing network functions and their life cycles. This paper proposes a software defined IoT architecture that captures the spirit of SDN and NFV where a software-defined Internet of Things (SD-IoT) controller can provide services as requested by an application and also manage heterogeneous physical sensors through their virtual representation called software-defined virtual sensor (SD-VSensor) autonomously. In particular, the paper presents the design of a streamline SD-IoT controller, a lightweight and reconfigurable SD-VSensor, and the communication protocol (S-MANAGE) between them. The proposed architecture enables heterogeneous application-specific WSN systems to be recognized and effectively utilized by diverse IoT applications under the orchestration of the SD-IoT controller. Moreover, heterogeneity of sensor nodes or IoT devices can be programmed to achieve sensor services on demand. The preliminary implementation results demonstrate the feasibility and efficiency of the proposed architecture.